From 75ced83574ae11fac3a54db2993d5aed0b42c8bc Mon Sep 17 00:00:00 2001
From: wassname <1103714+wassname@users.noreply.github.com>
Date: Mon, 6 Jan 2025 12:12:19 +0800
Subject: [PATCH] wip

---
 bs_writing_detector/data/load_md.py           |   22 +
 bs_writing_detector/metrics/ppx.py            |   44 +
 nbs/01_detection_using_adapter_ft_split.ipynb | 3632 +++++++++++++----
 nbs/02_detection_using_tldr_prompt.ipynb      | 1805 +++++++-
 nbs/03_get_lesswrong_data.ipynb               |   28 +-
 5 files changed, 4524 insertions(+), 1007 deletions(-)
 create mode 100644 bs_writing_detector/data/load_md.py

diff --git a/bs_writing_detector/data/load_md.py b/bs_writing_detector/data/load_md.py
new file mode 100644
index 0000000..a479019
--- /dev/null
+++ b/bs_writing_detector/data/load_md.py
@@ -0,0 +1,22 @@
+import frontmatter
+from pathlib import Path
+from loguru import logger
+import pandas as pd
+
+def load_md(f: Path = Path("../samples/"), max_chars=2000):
+    sample_files = sorted(f.glob('*.md'))
+    samples = []
+    for f in sample_files:
+        logger.debug(f)
+        with open(f, "r") as file:
+            post = frontmatter.load(file)
+            samples.append(dict(content=post.content[:max_chars], f=f, **post.metadata))
+    return samples
+
+def load_md_df(f, max_chars=2000):
+    samples = load_md(f, max_chars=max_chars)
+    df = pd.DataFrame(samples)
+    df = df[['title', 'f', 'content', 'url', 'novelty', 'date']]
+    df['date'] = pd.to_datetime(df['date'], utc=True)
+    df['in_training'] = df.date < '2024-01-01'
+    return df
diff --git a/bs_writing_detector/metrics/ppx.py b/bs_writing_detector/metrics/ppx.py
index 3c4b732..2a2c057 100644
--- a/bs_writing_detector/metrics/ppx.py
+++ b/bs_writing_detector/metrics/ppx.py
@@ -94,6 +94,7 @@ def perplexity_compute(
             (loss_fct(shift_logits.transpose(1, 2), shift_labels) * shift_attention_mask_batch)
         )
         # remove all the masked ones
+        assert nll_batch.size(0) == 1
         nll_batch = nll_batch[shift_attention_mask_batch == 1][None, :] # FIXME only for batch_size=1
         perplexity_batch = torch.exp(nll_batch.mean(1)).cpu().numpy()
 
@@ -101,3 +102,46 @@ def perplexity_compute(
         nlls += nll_batch.cpu().numpy().tolist()
 
     return {"perplexities": np.array(ppls), "nlls": np.array(nlls)}
+
+
+# modified from https://github.dev/huggingface/evaluate/blob/8dfe05784099fb9af55b8e77793205a3b7c86465/measurements/perplexity/perplexity.py#L154
+import evaluate
+from evaluate import logging
+from torch.nn import CrossEntropyLoss
+from torch.utils.data import DataLoader
+
+def perplexity_compute_ds(
+    ds, model, tokenizer, batch_size: int = 16, add_start_token: bool = True, device=None, max_length=None
+):
+    model = model.to(device)
+
+
+    ds = ds.with_format('pt')
+    dl = DataLoader(ds, batch_size=1, shuffle=False, num_workers=0, collate_fn=tokenizer.pad, pin_memory=True)
+
+    ppls = []
+    nlls = []
+    loss_fct = CrossEntropyLoss(reduction="none")
+    for b in dl:
+        input_ids = b['input_ids'].to(device)
+        attention_mask = b['attention_mask'].to(device)
+
+        labels = input_ids
+
+        with torch.no_grad():
+            out_logits = model(input_ids=input_ids, attention_mask=attention_mask).logits
+
+        shift_logits = out_logits[..., :-1, :].contiguous()
+        shift_labels = labels[..., 1:].contiguous()
+        shift_attention_mask_batch = attention_mask[..., 1:].contiguous()
+
+        nll_batch = loss_fct(shift_logits.transpose(1, 2), shift_labels) * shift_attention_mask_batch
+        assert nll_batch.size(0) == 1
+        nll_batch = nll_batch[shift_attention_mask_batch == 1][None, :] # FIXME only for batch_size=1
+
+        perplexity_batch = torch.exp(nll_batch.sum(1) / shift_attention_mask_batch.sum(1)).cpu().numpy()
+
+        ppls += perplexity_batch.tolist()
+        nlls += nll_batch.cpu().numpy().tolist()
+
+    return {"perplexities": torch.tensor(ppls), "mean_perplexity": torch.tensor(ppls).mean(), "nlls": torch.tensor(nlls)}
diff --git a/nbs/01_detection_using_adapter_ft_split.ipynb b/nbs/01_detection_using_adapter_ft_split.ipynb
index f857b38..e88e580 100644
--- a/nbs/01_detection_using_adapter_ft_split.ipynb
+++ b/nbs/01_detection_using_adapter_ft_split.ipynb
@@ -11,16 +11,17 @@
    "cell_type": "code",
    "execution_count": 1,
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
-      "  from .autonotebook import tqdm as notebook_tqdm\n"
-     ]
-    }
-   ],
+   "outputs": [],
+   "source": [
+    "%reload_ext autoreload\n",
+    "%autoreload 2"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
    "source": [
     "from torch import optim\n",
     "import lightning as pl\n",
@@ -29,7 +30,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 2,
+   "execution_count": 3,
    "metadata": {},
    "outputs": [
     {
@@ -38,7 +39,7 @@
        "1"
       ]
      },
-     "execution_count": 2,
+     "execution_count": 3,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -54,7 +55,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 4,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -82,7 +83,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -95,21 +96,19 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 6,
    "metadata": {},
    "outputs": [],
    "source": [
-    "\n"
+    "max_chars = 2000\n"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 7,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# model_name = \"TheBloke/phi-2-GPTQ\"\n",
-    "model_name = \"microsoft/phi-2\"\n",
     "\n",
     "# https://huggingface.co/collections/unsloth/llama-32-66f46afde4ca573864321a22\n",
     "model_name = \"unsloth/Llama-3.2-1B\"\n",
@@ -151,15 +150,15 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
+   "execution_count": 8,
    "metadata": {},
    "outputs": [
     {
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "Loading checkpoint shards: 100%|██████████| 2/2 [00:01<00:00,  1.77it/s]\n",
-      "Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.\n"
+      "Unused kwargs: ['_load_in_4bit', '_load_in_8bit', 'quant_method']. These kwargs are not used in <class 'transformers.utils.quantization_config.BitsAndBytesConfig'>.\n",
+      "`low_cpu_mem_usage` was None, now default to True since model is quantized.\n"
      ]
     }
    ],
@@ -171,7 +170,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 9,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -193,119 +192,24 @@
     "        #     feedforward_modules=[\"fc2\", 'fc1', 'out_proj'],\n",
     "        #     inference_mode=False,\n",
     "    )\n",
-    "    model = get_peft_model(base_model, peft_config)\n",
+    "    random_name = \"peft_\" + str(np.random.randint(0, 100000))\n",
+    "    model = get_peft_model(base_model, peft_config, adapter_name=random_name)\n",
     "    model.config.use_cache = False\n",
     "    return model\n",
     "\n",
     "model = reset_model(base_model)"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": 8,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "[PosixPath('../samples/bletchley_decleration.md'), PosixPath('../samples/cicero_fin1.md'), PosixPath('../samples/disney_appointment.md'), PosixPath('../samples/fake_paper.md'), PosixPath('../samples/fauci_emails.md'), PosixPath('../samples/harvard_announcement_reminders.md'), PosixPath('../samples/how_to_catch_a_liar.md'), PosixPath('../samples/lk-99_end.md'), PosixPath('../samples/lk-99_espanol.md'), PosixPath('../samples/lorem_ipsum.md'), PosixPath('../samples/openai_board_ann.md'), PosixPath('../samples/openai_paper_weak_to_strong.md'), PosixPath('../samples/politics_is_the_mind_killer.md'), PosixPath('../samples/statement_vyKamala_on_passing_of_johnson.md'), PosixPath('../samples/survey_of_rumours.md')]\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "dict_keys(['f', 'title', 'url', 'content'])"
-      ]
-     },
-     "execution_count": 8,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "\n",
-    "MAX_LEN = 400\n",
-    "\n",
-    "import frontmatter\n",
-    "from pathlib import Path\n",
-    "sample_files = sorted(Path(\"../samples/\").glob('*.md'))\n",
-    "print(sample_files)\n",
-    "samples = [{'f':f, **frontmatter.load(f).to_dict()} for f in sample_files]\n",
-    "\n",
-    "for sample in samples:\n",
-    "    assert 'title' in sample, sample['f']\n",
-    "    assert 'content' in sample\n",
-    "samples[0].keys()"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "## Helpers"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 9,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# modified from https://github.dev/huggingface/evaluate/blob/8dfe05784099fb9af55b8e77793205a3b7c86465/measurements/perplexity/perplexity.py#L154\n",
-    "import evaluate\n",
-    "from evaluate import logging\n",
-    "from torch.nn import CrossEntropyLoss\n",
-    "from torch.utils.data import DataLoader\n",
-    "\n",
-    "def perplexity_compute(\n",
-    "    ds, model, tokenizer, batch_size: int = 16, add_start_token: bool = True, device=None, max_length=None\n",
-    "):\n",
-    "    model = model.to(device)\n",
-    "\n",
-    "\n",
-    "    ds = ds.with_format('pt')\n",
-    "    dl = DataLoader(ds, batch_size=1, shuffle=False, num_workers=0, collate_fn=tokenizer.pad, pin_memory=True)\n",
-    "    ppls = []\n",
-    "    loss_fct = CrossEntropyLoss(reduction=\"none\")\n",
-    "    for b in dl:\n",
-    "        input_ids = b['input_ids'].to(device)\n",
-    "        attention_mask = b['attention_mask'].to(device)\n",
-    "        # print(attention_mask)\n",
-    "\n",
-    "        labels = input_ids\n",
-    "\n",
-    "        with torch.no_grad():\n",
-    "            out_logits = model(input_ids=input_ids, attention_mask=attention_mask).logits\n",
-    "\n",
-    "        shift_logits = out_logits[..., :-1, :].contiguous()\n",
-    "        shift_labels = labels[..., 1:].contiguous()\n",
-    "        shift_attention_mask_batch = attention_mask[..., 1:].contiguous()\n",
-    "\n",
-    "        perplexity_batch = torch.exp(\n",
-    "            (loss_fct(shift_logits.transpose(1, 2), shift_labels) * shift_attention_mask_batch).sum(1)\n",
-    "            / shift_attention_mask_batch.sum(1)\n",
-    "        )\n",
-    "\n",
-    "        ppls += perplexity_batch.tolist()\n",
-    "\n",
-    "    return {\"perplexities\": ppls, \"mean_perplexity\": torch.tensor(ppls).mean()}"
-   ]
-  },
   {
    "cell_type": "code",
    "execution_count": 10,
    "metadata": {},
    "outputs": [],
    "source": [
-    "# perplexity_compute(ds=ds_val, model=model, tokenizer=tokenizer, device='cuda')"
+    "from bs_writing_detector.data.load_md import load_md_df\n",
+    "from bs_writing_detector.metrics.ppx import perplexity_compute_ds\n",
+    "from pathlib import Path\n",
+    "df = load_md_df(Path(\"../samples/\"))"
    ]
   },
   {
@@ -333,6 +237,13 @@
     "## Lightning helpers"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
   {
    "cell_type": "code",
    "execution_count": 12,
@@ -343,9 +254,9 @@
     "    model.eval();\n",
     "    with torch.no_grad():\n",
     "        with model.disable_adapter():\n",
-    "            results = perplexity_compute(ds=ds_val, model=model, tokenizer=tokenizer, device='cuda')\n",
-    "        results2 = perplexity_compute(ds=ds_val, model=model, tokenizer=tokenizer, device='cuda')\n",
-    "    return dict(before=results['mean_perplexity'].item(), after=results2['mean_perplexity'].item())\n"
+    "            results = perplexity_compute_ds(ds=ds_val, model=model, tokenizer=tokenizer, device='cuda')['nlls'][0]\n",
+    "        results2 = perplexity_compute_ds(ds=ds_val, model=model, tokenizer=tokenizer, device='cuda')['nlls'][0]\n",
+    "    return dict(before=results, after=results2)\n"
    ]
   },
   {
@@ -382,43 +293,16 @@
    "execution_count": 14,
    "metadata": {},
    "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "(400, 14201)"
-      ]
-     },
-     "execution_count": 14,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "MAX_LEN, len(sample['content'])//3"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 15,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "Token indices sequence length is longer than the specified maximum sequence length for this model (8172 > 2048). Running this sequence through the model will result in indexing errors\n"
-     ]
-    },
     {
      "data": {
       "text/plain": [
        "Dataset({\n",
        "    features: ['input_ids', 'attention_mask', 'overflow_to_sample_mapping'],\n",
-       "    num_rows: 11\n",
+       "    num_rows: 2\n",
        "})"
       ]
      },
-     "execution_count": 15,
+     "execution_count": 14,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -428,7 +312,7 @@
     "\n",
     "def tokenize_and_split(examples):\n",
     "    l = len(tokenizer(examples).input_ids[0])\n",
-    "    max_len = min(l//3, MAX_LEN) # break into at least 5\n",
+    "    max_len = min(l//3, max_chars) # break into at least 5\n",
     "    max_len = max(max_len, 10)\n",
     "\n",
     "\n",
@@ -436,13 +320,14 @@
     "        examples,\n",
     "        add_special_tokens=False,\n",
     "        truncation=True,\n",
-    "        stride=2,\n",
+    "        stride=2, # – If set to a number along with max_length, the overflowing tokens returned will contain some tokens from the main sequence returned. The value of this argument defines the number of additional tokens.\n",
     "        max_length=max_len,\n",
     "        return_overflowing_tokens=True,\n",
     "        return_attention_mask=True,\n",
     "    )\n",
     "    return result\n",
     "\n",
+    "sample = df.sample(1).iloc[0]\n",
     "s = sample['content']\n",
     "d = Dataset.from_dict(tokenize_and_split([s]))\n",
     "d2  = d.train_test_split(test_size=0.5, seed=42)\n",
@@ -460,7 +345,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 16,
+   "execution_count": 15,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -479,8 +364,13 @@
     "    ds_train = d2['train']\n",
     "    ds_val = d2['test']\n",
     "\n",
-    "    \n",
+    "    print(model.peft_config)\n",
     "    model = reset_model(base_model)\n",
+    "\n",
+    "    # verify that we have reset it\n",
+    "    print(model.peft_config)\n",
+    "\n",
+    "\n",
     "    # eval(model, tokenizer, ds_train)\n",
     "\n",
     "    # https://huggingface.co/docs/transformers/v4.36.1/en/main_classes/trainer#transformers.Trainer\n",
@@ -537,411 +427,22 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 17,
+   "execution_count": 42,
    "metadata": {},
    "outputs": [
     {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      " 17%|█▋        | 1/6 [00:00<00:02,  2.35it/s]"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "{'loss': 2.8049, 'learning_rate': 0.001, 'epoch': 0.5}\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      " 33%|███▎      | 2/6 [00:00<00:01,  2.80it/s]"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "{'loss': 2.3359, 'learning_rate': 0.0008, 'epoch': 1.0}\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "                                             \n",
-      " 33%|███▎      | 2/6 [00:00<00:01,  2.80it/s]"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "{'eval_loss': 2.8958373069763184, 'eval_runtime': 0.2605, 'eval_samples_per_second': 7.677, 'eval_steps_per_second': 3.839, 'epoch': 1.0}\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      " 50%|█████     | 3/6 [00:01<00:01,  2.08it/s]"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "{'loss': 2.7644, 'learning_rate': 0.0006, 'epoch': 1.5}\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      " 67%|██████▋   | 4/6 [00:01<00:00,  2.42it/s]"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "{'loss': 2.297, 'learning_rate': 0.0004, 'epoch': 2.0}\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "                                             \n",
-      " 67%|██████▋   | 4/6 [00:01<00:00,  2.42it/s]"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "{'eval_loss': 2.8877780437469482, 'eval_runtime': 0.2592, 'eval_samples_per_second': 7.715, 'eval_steps_per_second': 3.857, 'epoch': 2.0}\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      " 83%|████████▎ | 5/6 [00:02<00:00,  2.05it/s]"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "{'loss': 2.2516, 'learning_rate': 0.0002, 'epoch': 2.5}\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "100%|██████████| 6/6 [00:02<00:00,  2.35it/s]"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "{'loss': 2.6997, 'learning_rate': 0.0, 'epoch': 3.0}\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "                                             \n",
-      "100%|██████████| 6/6 [00:02<00:00,  2.05it/s]\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "{'eval_loss': 2.884704113006592, 'eval_runtime': 0.2598, 'eval_samples_per_second': 7.699, 'eval_steps_per_second': 3.85, 'epoch': 3.0}\n",
-      "{'train_runtime': 2.9316, 'train_samples_per_second': 2.047, 'train_steps_per_second': 2.047, 'train_loss': 2.525577942530314, 'epoch': 3.0}\n",
-      "blechley declaration\n",
-      "{'before': 17.869091033935547, 'after': 17.52420425415039}\n",
-      "{'loss': 4.625, 'learning_rate': 0.0005, 'epoch': 0.17}\n",
-      "{'loss': 4.5422, 'learning_rate': 0.001, 'epoch': 0.33}\n",
-      "{'loss': 4.5928, 'learning_rate': 0.0009375, 'epoch': 0.5}\n",
-      "{'loss': 4.5384, 'learning_rate': 0.000875, 'epoch': 0.67}\n",
-      "{'loss': 4.7047, 'learning_rate': 0.0008125000000000001, 'epoch': 0.83}\n",
-      "{'loss': 4.4394, 'learning_rate': 0.00075, 'epoch': 1.0}\n",
-      "{'eval_loss': 4.234132289886475, 'eval_runtime': 0.8515, 'eval_samples_per_second': 7.047, 'eval_steps_per_second': 1.174, 'epoch': 1.0}\n",
-      "{'loss': 4.4981, 'learning_rate': 0.0006875, 'epoch': 1.17}\n",
-      "{'loss': 4.5187, 'learning_rate': 0.000625, 'epoch': 1.33}\n",
-      "{'loss': 4.419, 'learning_rate': 0.0005625000000000001, 'epoch': 1.5}\n",
-      "{'loss': 4.6345, 'learning_rate': 0.0005, 'epoch': 1.67}\n",
-      "{'loss': 4.4387, 'learning_rate': 0.0004375, 'epoch': 1.83}\n",
-      "{'loss': 4.3689, 'learning_rate': 0.000375, 'epoch': 2.0}\n",
-      "{'eval_loss': 4.212051868438721, 'eval_runtime': 0.7417, 'eval_samples_per_second': 8.09, 'eval_steps_per_second': 1.348, 'epoch': 2.0}\n",
-      "{'loss': 4.3709, 'learning_rate': 0.0003125, 'epoch': 2.17}\n",
-      "{'loss': 4.4461, 'learning_rate': 0.00025, 'epoch': 2.33}\n",
-      "{'loss': 4.5014, 'learning_rate': 0.0001875, 'epoch': 2.5}\n",
-      "{'loss': 4.5928, 'learning_rate': 0.000125, 'epoch': 2.67}\n",
-      "{'loss': 4.4311, 'learning_rate': 6.25e-05, 'epoch': 2.83}\n",
-      "{'loss': 4.4222, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 4.2056355476379395, 'eval_runtime': 0.7622, 'eval_samples_per_second': 7.872, 'eval_steps_per_second': 1.312, 'epoch': 3.0}\n",
-      "{'train_runtime': 8.1788, 'train_samples_per_second': 2.201, 'train_steps_per_second': 2.201, 'train_loss': 4.504713243908352, 'epoch': 3.0}\n",
-      "cicero from ibois, Philippe (2012-06-03).\n",
-      "{'before': 72.58737182617188, 'after': 67.74419403076172}\n",
-      "{'loss': 2.187, 'learning_rate': 0.001, 'epoch': 0.5}\n",
-      "{'loss': 2.0009, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
-      "{'eval_loss': 1.9654667377471924, 'eval_runtime': 0.178, 'eval_samples_per_second': 11.238, 'eval_steps_per_second': 5.619, 'epoch': 1.0}\n",
-      "{'loss': 2.1424, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
-      "{'loss': 1.9167, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
-      "{'eval_loss': 1.9636929035186768, 'eval_runtime': 0.1788, 'eval_samples_per_second': 11.188, 'eval_steps_per_second': 5.594, 'epoch': 2.0}\n",
-      "{'loss': 1.8877, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
-      "{'loss': 2.0672, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 1.9641517400741577, 'eval_runtime': 0.184, 'eval_samples_per_second': 10.869, 'eval_steps_per_second': 5.435, 'epoch': 3.0}\n",
-      "{'train_runtime': 1.978, 'train_samples_per_second': 3.033, 'train_steps_per_second': 3.033, 'train_loss': 2.033644914627075, 'epoch': 3.0}\n",
-      "disney appointment\n",
-      "{'before': 118.82615661621094, 'after': 118.42012786865234}\n",
-      "{'loss': 2.0357, 'learning_rate': 0.001, 'epoch': 0.5}\n",
-      "{'loss': 1.9873, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
-      "{'eval_loss': 1.807045817375183, 'eval_runtime': 0.1406, 'eval_samples_per_second': 14.221, 'eval_steps_per_second': 7.11, 'epoch': 1.0}\n",
-      "{'loss': 2.0176, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
-      "{'loss': 1.9243, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
-      "{'eval_loss': 1.7954585552215576, 'eval_runtime': 0.1396, 'eval_samples_per_second': 14.328, 'eval_steps_per_second': 7.164, 'epoch': 2.0}\n",
-      "{'loss': 1.9109, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
-      "{'loss': 1.9423, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 1.7919256687164307, 'eval_runtime': 0.14, 'eval_samples_per_second': 14.289, 'eval_steps_per_second': 7.145, 'epoch': 3.0}\n",
-      "{'train_runtime': 1.5666, 'train_samples_per_second': 3.83, 'train_steps_per_second': 3.83, 'train_loss': 1.969655990600586, 'epoch': 3.0}\n",
-      "fake ai hoax paper\n",
-      "{'before': 7.766979217529297, 'after': 7.697228908538818}\n",
-      "{'loss': 2.4331, 'learning_rate': 0.0005, 'epoch': 0.2}\n",
-      "{'loss': 2.9004, 'learning_rate': 0.001, 'epoch': 0.4}\n",
-      "{'loss': 3.5675, 'learning_rate': 0.0009230769230769232, 'epoch': 0.6}\n",
-      "{'loss': 3.3972, 'learning_rate': 0.0008461538461538462, 'epoch': 0.8}\n",
-      "{'loss': 3.5431, 'learning_rate': 0.0007692307692307693, 'epoch': 1.0}\n",
-      "{'eval_loss': 3.0353567600250244, 'eval_runtime': 0.8628, 'eval_samples_per_second': 6.954, 'eval_steps_per_second': 1.159, 'epoch': 1.0}\n",
-      "{'loss': 3.4732, 'learning_rate': 0.0006923076923076923, 'epoch': 1.2}\n",
-      "{'loss': 3.5319, 'learning_rate': 0.0006153846153846154, 'epoch': 1.4}\n",
-      "{'loss': 2.7887, 'learning_rate': 0.0005384615384615384, 'epoch': 1.6}\n",
-      "{'loss': 2.3222, 'learning_rate': 0.0004615384615384616, 'epoch': 1.8}\n",
-      "{'loss': 3.3527, 'learning_rate': 0.00038461538461538467, 'epoch': 2.0}\n",
-      "{'eval_loss': 3.0194509029388428, 'eval_runtime': 0.7424, 'eval_samples_per_second': 8.082, 'eval_steps_per_second': 1.347, 'epoch': 2.0}\n",
-      "{'loss': 2.2585, 'learning_rate': 0.0003076923076923077, 'epoch': 2.2}\n",
-      "{'loss': 3.3084, 'learning_rate': 0.0002307692307692308, 'epoch': 2.4}\n",
-      "{'loss': 3.4762, 'learning_rate': 0.00015384615384615385, 'epoch': 2.6}\n",
-      "{'loss': 3.2907, 'learning_rate': 7.692307692307693e-05, 'epoch': 2.8}\n",
-      "{'loss': 2.7338, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 3.014026641845703, 'eval_runtime': 0.7435, 'eval_samples_per_second': 8.07, 'eval_steps_per_second': 1.345, 'epoch': 3.0}\n",
-      "{'train_runtime': 7.2582, 'train_samples_per_second': 2.067, 'train_steps_per_second': 2.067, 'train_loss': 3.09183349609375, 'epoch': 3.0}\n",
-      "buzzfeed foi fauci emails 2023\n",
-      "{'before': 23.3094425201416, 'after': 22.406410217285156}\n",
-      "{'loss': 2.6779, 'learning_rate': 0.001, 'epoch': 0.5}\n",
-      "{'loss': 2.4325, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
-      "{'eval_loss': 2.7339022159576416, 'eval_runtime': 0.2193, 'eval_samples_per_second': 9.121, 'eval_steps_per_second': 4.561, 'epoch': 1.0}\n",
-      "{'loss': 2.6596, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
-      "{'loss': 2.3446, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
-      "{'eval_loss': 2.730224609375, 'eval_runtime': 0.2245, 'eval_samples_per_second': 8.91, 'eval_steps_per_second': 4.455, 'epoch': 2.0}\n",
-      "{'loss': 2.3111, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
-      "{'loss': 2.5791, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 2.729649543762207, 'eval_runtime': 0.2204, 'eval_samples_per_second': 9.076, 'eval_steps_per_second': 4.538, 'epoch': 3.0}\n",
-      "{'train_runtime': 2.2643, 'train_samples_per_second': 2.65, 'train_steps_per_second': 2.65, 'train_loss': 2.500802000363668, 'epoch': 3.0}\n",
-      "harvard announcment caplain israel hamas\n",
-      "{'before': 45.347354888916016, 'after': 45.12727355957031}\n",
-      "{'loss': 2.6223, 'learning_rate': 0.001, 'epoch': 0.5}\n",
-      "{'loss': 2.4621, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
-      "{'eval_loss': 3.127470016479492, 'eval_runtime': 0.378, 'eval_samples_per_second': 5.291, 'eval_steps_per_second': 2.645, 'epoch': 1.0}\n",
-      "{'loss': 2.5923, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
-      "{'loss': 2.4313, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
-      "{'eval_loss': 3.115180253982544, 'eval_runtime': 0.2642, 'eval_samples_per_second': 7.569, 'eval_steps_per_second': 3.785, 'epoch': 2.0}\n",
-      "{'loss': 2.3652, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
-      "{'loss': 2.5058, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 3.1118650436401367, 'eval_runtime': 0.2632, 'eval_samples_per_second': 7.6, 'eval_steps_per_second': 3.8, 'epoch': 3.0}\n",
-      "{'train_runtime': 2.9644, 'train_samples_per_second': 2.024, 'train_steps_per_second': 2.024, 'train_loss': 2.4964850743611655, 'epoch': 3.0}\n",
-      "How to Catch an AI Liar\n",
-      "{'before': 28.949859619140625, 'after': 28.0087890625}\n",
-      "{'loss': 3.5646, 'learning_rate': 0.0005, 'epoch': 0.2}\n",
-      "{'loss': 2.8252, 'learning_rate': 0.001, 'epoch': 0.4}\n",
-      "{'loss': 2.0116, 'learning_rate': 0.0009230769230769232, 'epoch': 0.6}\n",
-      "{'loss': 2.7551, 'learning_rate': 0.0008461538461538462, 'epoch': 0.8}\n",
-      "{'loss': 2.4442, 'learning_rate': 0.0007692307692307693, 'epoch': 1.0}\n",
-      "{'eval_loss': 2.635437488555908, 'eval_runtime': 0.7632, 'eval_samples_per_second': 6.552, 'eval_steps_per_second': 1.31, 'epoch': 1.0}\n",
-      "{'loss': 1.9103, 'learning_rate': 0.0006923076923076923, 'epoch': 1.2}\n",
-      "{'loss': 2.4336, 'learning_rate': 0.0006153846153846154, 'epoch': 1.4}\n",
-      "{'loss': 2.7349, 'learning_rate': 0.0005384615384615384, 'epoch': 1.6}\n",
-      "{'loss': 3.4216, 'learning_rate': 0.0004615384615384616, 'epoch': 1.8}\n",
-      "{'loss': 2.6993, 'learning_rate': 0.00038461538461538467, 'epoch': 2.0}\n",
-      "{'eval_loss': 2.623994827270508, 'eval_runtime': 0.6368, 'eval_samples_per_second': 7.852, 'eval_steps_per_second': 1.57, 'epoch': 2.0}\n",
-      "{'loss': 3.3393, 'learning_rate': 0.0003076923076923077, 'epoch': 2.2}\n",
-      "{'loss': 1.8571, 'learning_rate': 0.0002307692307692308, 'epoch': 2.4}\n",
-      "{'loss': 2.3943, 'learning_rate': 0.00015384615384615385, 'epoch': 2.6}\n",
-      "{'loss': 2.6463, 'learning_rate': 7.692307692307693e-05, 'epoch': 2.8}\n",
-      "{'loss': 2.6925, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 2.6198740005493164, 'eval_runtime': 0.636, 'eval_samples_per_second': 7.862, 'eval_steps_per_second': 1.572, 'epoch': 3.0}\n",
-      "{'train_runtime': 6.4124, 'train_samples_per_second': 2.339, 'train_steps_per_second': 2.339, 'train_loss': 2.648661724726359, 'epoch': 3.0}\n",
-      "LK-99-en\n",
-      "{'before': 14.513845443725586, 'after': 14.066093444824219}\n",
-      "{'loss': 2.9979, 'learning_rate': 0.0005, 'epoch': 0.17}\n",
-      "{'loss': 2.0904, 'learning_rate': 0.001, 'epoch': 0.33}\n",
-      "{'loss': 2.7862, 'learning_rate': 0.0009375, 'epoch': 0.5}\n",
-      "{'loss': 2.4246, 'learning_rate': 0.000875, 'epoch': 0.67}\n",
-      "{'loss': 2.6095, 'learning_rate': 0.0008125000000000001, 'epoch': 0.83}\n",
-      "{'loss': 2.5655, 'learning_rate': 0.00075, 'epoch': 1.0}\n",
-      "{'eval_loss': 2.335665464401245, 'eval_runtime': 0.9768, 'eval_samples_per_second': 7.166, 'eval_steps_per_second': 1.024, 'epoch': 1.0}\n",
-      "{'loss': 2.3562, 'learning_rate': 0.0006875, 'epoch': 1.17}\n",
-      "{'loss': 2.8196, 'learning_rate': 0.000625, 'epoch': 1.33}\n",
-      "{'loss': 2.5404, 'learning_rate': 0.0005625000000000001, 'epoch': 1.5}\n",
-      "{'loss': 2.4861, 'learning_rate': 0.0005, 'epoch': 1.67}\n",
-      "{'loss': 2.6823, 'learning_rate': 0.0004375, 'epoch': 1.83}\n",
-      "{'loss': 2.0126, 'learning_rate': 0.000375, 'epoch': 2.0}\n",
-      "{'eval_loss': 2.320927858352661, 'eval_runtime': 0.867, 'eval_samples_per_second': 8.074, 'eval_steps_per_second': 1.153, 'epoch': 2.0}\n",
-      "{'loss': 2.4835, 'learning_rate': 0.0003125, 'epoch': 2.17}\n",
-      "{'loss': 2.5887, 'learning_rate': 0.00025, 'epoch': 2.33}\n",
-      "{'loss': 2.7585, 'learning_rate': 0.0001875, 'epoch': 2.5}\n",
-      "{'loss': 2.4708, 'learning_rate': 0.000125, 'epoch': 2.67}\n",
-      "{'loss': 2.2034, 'learning_rate': 6.25e-05, 'epoch': 2.83}\n",
-      "{'loss': 1.9709, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 2.3152239322662354, 'eval_runtime': 0.8652, 'eval_samples_per_second': 8.091, 'eval_steps_per_second': 1.156, 'epoch': 3.0}\n",
-      "{'train_runtime': 8.6099, 'train_samples_per_second': 2.091, 'train_steps_per_second': 2.091, 'train_loss': 2.4914957218699985, 'epoch': 3.0}\n",
-      "LK-99-es\n",
-      "{'before': 11.415018081665039, 'after': 10.97294807434082}\n",
-      "{'loss': 1.8845, 'learning_rate': 0.0003333333333333333, 'epoch': 0.11}\n",
-      "{'loss': 1.9495, 'learning_rate': 0.0006666666666666666, 'epoch': 0.22}\n",
-      "{'loss': 1.9686, 'learning_rate': 0.001, 'epoch': 0.33}\n",
-      "{'loss': 1.9168, 'learning_rate': 0.0009583333333333334, 'epoch': 0.44}\n",
-      "{'loss': 1.9844, 'learning_rate': 0.0009166666666666666, 'epoch': 0.56}\n",
-      "{'loss': 1.9139, 'learning_rate': 0.000875, 'epoch': 0.67}\n",
-      "{'loss': 1.8329, 'learning_rate': 0.0008333333333333334, 'epoch': 0.78}\n",
-      "{'loss': 1.8556, 'learning_rate': 0.0007916666666666666, 'epoch': 0.89}\n",
-      "{'loss': 1.8869, 'learning_rate': 0.0007916666666666666, 'epoch': 1.0}\n",
-      "{'eval_loss': 1.836249828338623, 'eval_runtime': 1.2543, 'eval_samples_per_second': 7.175, 'eval_steps_per_second': 1.595, 'epoch': 1.0}\n",
-      "{'loss': 1.8902, 'learning_rate': 0.00075, 'epoch': 1.11}\n",
-      "{'loss': 1.8931, 'learning_rate': 0.0007083333333333334, 'epoch': 1.22}\n",
-      "{'loss': 1.8021, 'learning_rate': 0.0006666666666666666, 'epoch': 1.33}\n",
-      "{'loss': 1.884, 'learning_rate': 0.000625, 'epoch': 1.44}\n",
-      "{'loss': 1.7923, 'learning_rate': 0.0005833333333333334, 'epoch': 1.56}\n",
-      "{'loss': 1.848, 'learning_rate': 0.0005416666666666666, 'epoch': 1.67}\n",
-      "{'loss': 1.8103, 'learning_rate': 0.0005, 'epoch': 1.78}\n",
-      "{'loss': 1.9024, 'learning_rate': 0.0004583333333333333, 'epoch': 1.89}\n",
-      "{'loss': 1.9309, 'learning_rate': 0.0004166666666666667, 'epoch': 2.0}\n",
-      "{'eval_loss': 1.829906702041626, 'eval_runtime': 1.1467, 'eval_samples_per_second': 7.849, 'eval_steps_per_second': 1.744, 'epoch': 2.0}\n",
-      "{'loss': 1.9011, 'learning_rate': 0.000375, 'epoch': 2.11}\n",
-      "{'loss': 1.8506, 'learning_rate': 0.0003333333333333333, 'epoch': 2.22}\n",
-      "{'loss': 1.8036, 'learning_rate': 0.0002916666666666667, 'epoch': 2.33}\n",
-      "{'loss': 1.828, 'learning_rate': 0.00025, 'epoch': 2.44}\n",
-      "{'loss': 1.9094, 'learning_rate': 0.00020833333333333335, 'epoch': 2.56}\n",
-      "{'loss': 1.8678, 'learning_rate': 0.00016666666666666666, 'epoch': 2.67}\n",
-      "{'loss': 1.8555, 'learning_rate': 0.000125, 'epoch': 2.78}\n",
-      "{'loss': 1.7289, 'learning_rate': 8.333333333333333e-05, 'epoch': 2.89}\n",
-      "{'loss': 1.7859, 'learning_rate': 4.1666666666666665e-05, 'epoch': 3.0}\n",
-      "{'eval_loss': 1.8297154903411865, 'eval_runtime': 1.1477, 'eval_samples_per_second': 7.841, 'eval_steps_per_second': 1.743, 'epoch': 3.0}\n",
-      "{'train_runtime': 12.5295, 'train_samples_per_second': 2.155, 'train_steps_per_second': 2.155, 'train_loss': 1.8695231985162806, 'epoch': 3.0}\n",
-      "Lorem ipsum\n",
-      "{'before': 6.564840316772461, 'after': 6.268791675567627}\n",
-      "{'loss': 2.2808, 'learning_rate': 0.001, 'epoch': 0.5}\n",
-      "{'loss': 2.264, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
-      "{'eval_loss': 2.300767183303833, 'eval_runtime': 0.1412, 'eval_samples_per_second': 14.167, 'eval_steps_per_second': 7.084, 'epoch': 1.0}\n",
-      "{'loss': 2.1983, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
-      "{'loss': 2.1913, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
-      "{'eval_loss': 2.2839250564575195, 'eval_runtime': 0.1403, 'eval_samples_per_second': 14.254, 'eval_steps_per_second': 7.127, 'epoch': 2.0}\n",
-      "{'loss': 2.149, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
-      "{'loss': 2.095, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 2.2791788578033447, 'eval_runtime': 0.1407, 'eval_samples_per_second': 14.211, 'eval_steps_per_second': 7.105, 'epoch': 3.0}\n",
-      "{'train_runtime': 1.5544, 'train_samples_per_second': 3.86, 'train_steps_per_second': 3.86, 'train_loss': 2.196397304534912, 'epoch': 3.0}\n",
-      "openai board ann\n",
-      "{'before': 55.808502197265625, 'after': 54.66792297363281}\n",
-      "{'loss': 2.8464, 'learning_rate': 0.001, 'epoch': 0.5}\n",
-      "{'loss': 2.4796, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
-      "{'eval_loss': 2.501676082611084, 'eval_runtime': 0.3861, 'eval_samples_per_second': 5.18, 'eval_steps_per_second': 2.59, 'epoch': 1.0}\n",
-      "{'loss': 2.8062, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
-      "{'loss': 2.4146, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
-      "{'eval_loss': 2.4917240142822266, 'eval_runtime': 0.2752, 'eval_samples_per_second': 7.268, 'eval_steps_per_second': 3.634, 'epoch': 2.0}\n",
-      "{'loss': 2.356, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
-      "{'loss': 2.7228, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 2.488405704498291, 'eval_runtime': 0.2661, 'eval_samples_per_second': 7.515, 'eval_steps_per_second': 3.757, 'epoch': 3.0}\n",
-      "{'train_runtime': 3.0083, 'train_samples_per_second': 1.994, 'train_steps_per_second': 1.994, 'train_loss': 2.60427188873291, 'epoch': 3.0}\n",
-      "weak to strong\n",
-      "{'before': 46.764183044433594, 'after': 46.40473556518555}\n",
-      "{'loss': 2.7259, 'learning_rate': 0.001, 'epoch': 0.5}\n",
-      "{'loss': 2.7937, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
-      "{'eval_loss': 2.7441396713256836, 'eval_runtime': 0.1515, 'eval_samples_per_second': 13.205, 'eval_steps_per_second': 6.603, 'epoch': 1.0}\n",
-      "{'loss': 2.6661, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
-      "{'loss': 2.6998, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
-      "{'eval_loss': 2.74137282371521, 'eval_runtime': 0.1514, 'eval_samples_per_second': 13.211, 'eval_steps_per_second': 6.605, 'epoch': 2.0}\n",
-      "{'loss': 2.6949, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
-      "{'loss': 2.5599, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 2.7393088340759277, 'eval_runtime': 0.1558, 'eval_samples_per_second': 12.833, 'eval_steps_per_second': 6.416, 'epoch': 3.0}\n",
-      "{'train_runtime': 1.6635, 'train_samples_per_second': 3.607, 'train_steps_per_second': 3.607, 'train_loss': 2.690062483151754, 'epoch': 3.0}\n",
-      "politics is the mind-killer\n",
-      "{'before': 247.5515594482422, 'after': 245.82681274414062}\n",
-      "{'loss': 1.8812, 'learning_rate': 0.001, 'epoch': 0.5}\n",
-      "{'loss': 2.2442, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
-      "{'eval_loss': 2.4278409481048584, 'eval_runtime': 0.0735, 'eval_samples_per_second': 27.196, 'eval_steps_per_second': 13.598, 'epoch': 1.0}\n",
-      "{'loss': 1.8079, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
-      "{'loss': 2.1305, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
-      "{'eval_loss': 2.4188129901885986, 'eval_runtime': 0.0764, 'eval_samples_per_second': 26.194, 'eval_steps_per_second': 13.097, 'epoch': 2.0}\n",
-      "{'loss': 2.1037, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
-      "{'loss': 1.6808, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 2.4169728755950928, 'eval_runtime': 0.0738, 'eval_samples_per_second': 27.114, 'eval_steps_per_second': 13.557, 'epoch': 3.0}\n",
-      "{'train_runtime': 0.9463, 'train_samples_per_second': 6.341, 'train_steps_per_second': 6.341, 'train_loss': 1.9747143586476643, 'epoch': 3.0}\n",
-      "statement by whitehouse on passing\n",
-      "{'before': 29.197132110595703, 'after': 28.9396915435791}\n",
-      "{'loss': 2.5398, 'learning_rate': 0.0003333333333333333, 'epoch': 0.1}\n",
-      "{'loss': 2.563, 'learning_rate': 0.0006666666666666666, 'epoch': 0.2}\n",
-      "{'loss': 2.362, 'learning_rate': 0.001, 'epoch': 0.3}\n",
-      "{'loss': 2.4908, 'learning_rate': 0.0009629629629629629, 'epoch': 0.4}\n",
-      "{'loss': 2.8433, 'learning_rate': 0.000925925925925926, 'epoch': 0.5}\n",
-      "{'loss': 3.1541, 'learning_rate': 0.0008888888888888888, 'epoch': 0.6}\n",
-      "{'loss': 2.9383, 'learning_rate': 0.0008518518518518519, 'epoch': 0.7}\n",
-      "{'loss': 2.6576, 'learning_rate': 0.0008148148148148148, 'epoch': 0.8}\n",
-      "{'loss': 2.6327, 'learning_rate': 0.0007777777777777778, 'epoch': 0.9}\n",
-      "{'loss': 2.6943, 'learning_rate': 0.0007407407407407407, 'epoch': 1.0}\n",
-      "{'eval_loss': 2.42354416847229, 'eval_runtime': 1.5423, 'eval_samples_per_second': 7.132, 'eval_steps_per_second': 1.297, 'epoch': 1.0}\n",
-      "{'loss': 2.3882, 'learning_rate': 0.0007037037037037038, 'epoch': 1.1}\n",
-      "{'loss': 2.738, 'learning_rate': 0.0006666666666666666, 'epoch': 1.2}\n",
-      "{'loss': 2.5917, 'learning_rate': 0.0006296296296296296, 'epoch': 1.3}\n",
-      "{'loss': 2.8582, 'learning_rate': 0.0005925925925925926, 'epoch': 1.4}\n",
-      "{'loss': 3.0774, 'learning_rate': 0.0005555555555555556, 'epoch': 1.5}\n",
-      "{'loss': 2.266, 'learning_rate': 0.0005185185185185185, 'epoch': 1.6}\n",
-      "{'loss': 2.4382, 'learning_rate': 0.00048148148148148144, 'epoch': 1.7}\n",
-      "{'loss': 2.6197, 'learning_rate': 0.0004444444444444444, 'epoch': 1.8}\n",
-      "{'loss': 2.3982, 'learning_rate': 0.0004074074074074074, 'epoch': 1.9}\n",
-      "{'loss': 2.5713, 'learning_rate': 0.00037037037037037035, 'epoch': 2.0}\n",
-      "{'eval_loss': 2.4057815074920654, 'eval_runtime': 1.6782, 'eval_samples_per_second': 6.555, 'eval_steps_per_second': 1.192, 'epoch': 2.0}\n",
-      "{'loss': 2.6508, 'learning_rate': 0.0003333333333333333, 'epoch': 2.1}\n",
-      "{'loss': 2.4209, 'learning_rate': 0.0002962962962962963, 'epoch': 2.2}\n",
-      "{'loss': 3.0109, 'learning_rate': 0.00025925925925925926, 'epoch': 2.3}\n",
-      "{'loss': 2.561, 'learning_rate': 0.0002222222222222222, 'epoch': 2.4}\n",
-      "{'loss': 2.6675, 'learning_rate': 0.00018518518518518518, 'epoch': 2.5}\n",
-      "{'loss': 2.4846, 'learning_rate': 0.00014814814814814815, 'epoch': 2.6}\n",
-      "{'loss': 2.7888, 'learning_rate': 0.0001111111111111111, 'epoch': 2.7}\n",
-      "{'loss': 2.327, 'learning_rate': 7.407407407407407e-05, 'epoch': 2.8}\n",
-      "{'loss': 2.3715, 'learning_rate': 3.7037037037037037e-05, 'epoch': 2.9}\n",
-      "{'loss': 2.2409, 'learning_rate': 0.0, 'epoch': 3.0}\n",
-      "{'eval_loss': 2.399984121322632, 'eval_runtime': 1.608, 'eval_samples_per_second': 6.841, 'eval_steps_per_second': 1.244, 'epoch': 3.0}\n",
-      "{'train_runtime': 15.0837, 'train_samples_per_second': 1.989, 'train_steps_per_second': 1.989, 'train_loss': 2.6115529775619506, 'epoch': 3.0}\n",
-      "Gemini to Q*\n",
-      "{'before': 11.756383895874023, 'after': 11.190644264221191}\n"
-     ]
+     "data": {
+      "text/plain": [
+       "dict_keys(['before', 'after', 'hist', 'train/before', 'train/after', 'title', 'f', 'content', 'url', 'novelty', 'date', 'in_training'])"
+      ]
+     },
+     "execution_count": 42,
+     "metadata": {},
+     "output_type": "execute_result"
     }
    ],
    "source": [
-    "data = []\n",
-    "for sample in samples:\n",
-    "    r = learn_sample(sample)\n",
-    "    print(sample['title'])\n",
-    "    print(dict(before=r['before'], after=r['after']))\n",
-    "    data.append(dict(**r, **sample))"
+    "data[0].keys()"
    ]
   },
   {
@@ -951,7 +452,817 @@
    "outputs": [
     {
      "data": {
-      "image/png": "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",
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "46ca413f7112493bbeb2be6403b68e9f",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/30 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 2.8911, 'grad_norm': 0.8076428174972534, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 2.3384, 'grad_norm': 0.6929064989089966, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.5387918949127197, 'eval_runtime': 0.1312, 'eval_samples_per_second': 15.24, 'eval_steps_per_second': 7.62, 'epoch': 1.0}\n",
+      "{'loss': 2.7436, 'grad_norm': 0.7699084281921387, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.1441, 'grad_norm': 0.6051006317138672, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.5293452739715576, 'eval_runtime': 0.1328, 'eval_samples_per_second': 15.055, 'eval_steps_per_second': 7.528, 'epoch': 2.0}\n",
+      "{'loss': 2.5154, 'grad_norm': 0.749704122543335, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.0386, 'grad_norm': 0.5838078856468201, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.5258588790893555, 'eval_runtime': 0.1318, 'eval_samples_per_second': 15.171, 'eval_steps_per_second': 7.585, 'epoch': 3.0}\n",
+      "{'train_runtime': 1.3375, 'train_samples_per_second': 4.486, 'train_steps_per_second': 4.486, 'train_loss': 2.445200721422831, 'epoch': 3.0}\n",
+      "Anthropic and Palantir Partner to Bring Claude AI Models to AWS for U.S. Government Intelligence and Defense Operations\n",
+      "{'before': [11.264070510864258, 7.382840156555176, 12.170625686645508, 1.2329556941986084], 'after': [11.19461441040039, 7.479499816894531, 12.069757461547852, 0.9895272254943848]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.2976, 'grad_norm': 1.1653801202774048, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.499, 'grad_norm': 1.3143912553787231, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.3701250553131104, 'eval_runtime': 0.1588, 'eval_samples_per_second': 12.593, 'eval_steps_per_second': 6.297, 'epoch': 1.0}\n",
+      "{'loss': 3.0943, 'grad_norm': 1.04639732837677, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 3.1374, 'grad_norm': 1.0442923307418823, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.351206064224243, 'eval_runtime': 0.127, 'eval_samples_per_second': 15.745, 'eval_steps_per_second': 7.872, 'epoch': 2.0}\n",
+      "{'loss': 2.7867, 'grad_norm': 0.9014344811439514, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.9493, 'grad_norm': 0.9459804892539978, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.344069242477417, 'eval_runtime': 0.1281, 'eval_samples_per_second': 15.616, 'eval_steps_per_second': 7.808, 'epoch': 3.0}\n",
+      "{'train_runtime': 1.3909, 'train_samples_per_second': 4.314, 'train_steps_per_second': 4.314, 'train_loss': 3.1273703972498574, 'epoch': 3.0}\n",
+      "TradingAgents: Multi-Agents LLM Financial Trading Framework\n",
+      "{'before': [10.053862571716309, 0.6636170744895935, 8.349092483520508, 1.452925443649292, 0.9928902387619019, 8.745048522949219], 'after': [9.95648193359375, 0.47155672311782837, 8.367955207824707, 1.2811455726623535, 0.875080943107605, 8.782474517822266]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.3063, 'grad_norm': 1.185476303100586, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.3628, 'grad_norm': 1.5031108856201172, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.6609110832214355, 'eval_runtime': 0.1593, 'eval_samples_per_second': 12.556, 'eval_steps_per_second': 6.278, 'epoch': 1.0}\n",
+      "{'loss': 3.0717, 'grad_norm': 0.9321146011352539, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 3.0218, 'grad_norm': 1.1784974336624146, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.6481850147247314, 'eval_runtime': 0.1593, 'eval_samples_per_second': 12.552, 'eval_steps_per_second': 6.276, 'epoch': 2.0}\n",
+      "{'loss': 2.7501, 'grad_norm': 0.7790453433990479, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.8441, 'grad_norm': 1.0859301090240479, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.6467812061309814, 'eval_runtime': 0.1589, 'eval_samples_per_second': 12.588, 'eval_steps_per_second': 6.294, 'epoch': 3.0}\n",
+      "{'train_runtime': 1.6521, 'train_samples_per_second': 3.632, 'train_steps_per_second': 3.632, 'train_loss': 3.059459924697876, 'epoch': 3.0}\n",
+      "Flower Crowns and Furry Mishaps by MyPalAI\n",
+      "{'before': [7.296568393707275, 3.3268048763275146, 3.3179941177368164, 4.614715576171875, 4.787782669067383, 6.660912990570068], 'after': [7.300139904022217, 3.366487979888916, 3.34420108795166, 4.758848190307617, 4.731474876403809, 6.640608787536621]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 4.1433, 'grad_norm': 1.3135420083999634, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.8979, 'grad_norm': 1.6989140510559082, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.8832738399505615, 'eval_runtime': 0.1567, 'eval_samples_per_second': 12.767, 'eval_steps_per_second': 6.384, 'epoch': 1.0}\n",
+      "{'loss': 3.8552, 'grad_norm': 1.1130166053771973, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 3.445, 'grad_norm': 1.2440484762191772, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.857309103012085, 'eval_runtime': 0.1665, 'eval_samples_per_second': 12.015, 'eval_steps_per_second': 6.007, 'epoch': 2.0}\n",
+      "{'loss': 3.437, 'grad_norm': 1.004622459411621, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 3.2143, 'grad_norm': 1.0474002361297607, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.846127986907959, 'eval_runtime': 0.1571, 'eval_samples_per_second': 12.729, 'eval_steps_per_second': 6.365, 'epoch': 3.0}\n",
+      "{'train_runtime': 2.0371, 'train_samples_per_second': 2.945, 'train_steps_per_second': 2.945, 'train_loss': 3.6654568910598755, 'epoch': 3.0}\n",
+      "Paradox's Box (Bobiverse) by Mark4man\n",
+      "{'before': [8.102655410766602, 4.443521976470947, 6.978631973266602, 3.8517813682556152, 6.285520076751709], 'after': [7.454442024230957, 4.559870719909668, 7.13922643661499, 4.0717997550964355, 6.418486595153809]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.5963, 'grad_norm': 1.7606483697891235, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.8611, 'grad_norm': nan, 'learning_rate': 0.001, 'epoch': 1.0}\n",
+      "{'eval_loss': 4.20759391784668, 'eval_runtime': 0.1417, 'eval_samples_per_second': 14.112, 'eval_steps_per_second': 7.056, 'epoch': 1.0}\n",
+      "{'loss': 3.5963, 'grad_norm': 1.7607665061950684, 'learning_rate': 0.0008, 'epoch': 1.5}\n",
+      "{'loss': 3.8058, 'grad_norm': 5.883133411407471, 'learning_rate': 0.0006, 'epoch': 2.0}\n",
+      "{'eval_loss': 4.105123519897461, 'eval_runtime': 0.1429, 'eval_samples_per_second': 13.994, 'eval_steps_per_second': 6.997, 'epoch': 2.0}\n",
+      "{'loss': 2.6548, 'grad_norm': 1.1668059825897217, 'learning_rate': 0.0004, 'epoch': 2.5}\n",
+      "{'loss': 3.4369, 'grad_norm': 1.433318853378296, 'learning_rate': 0.0002, 'epoch': 3.0}\n",
+      "{'eval_loss': 4.070194721221924, 'eval_runtime': 0.1445, 'eval_samples_per_second': 13.837, 'eval_steps_per_second': 6.918, 'epoch': 3.0}\n",
+      "{'train_runtime': 2.0652, 'train_samples_per_second': 2.905, 'train_steps_per_second': 2.905, 'train_loss': 3.4918715159098306, 'epoch': 3.0}\n",
+      "Deliberative Alignment: Reasoning Enables Safer Language Models\n",
+      "{'before': [11.071215629577637, 4.1971049308776855, 5.31649112701416, 0.9296282529830933], 'after': [10.824005126953125, 4.173127174377441, 5.1760783195495605, 1.0669198036193848]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 2.8892, 'grad_norm': 1.9384897947311401, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 2.5209, 'grad_norm': 1.7461295127868652, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 2.852980613708496, 'eval_runtime': 0.1324, 'eval_samples_per_second': 15.108, 'eval_steps_per_second': 7.554, 'epoch': 1.0}\n",
+      "{'loss': 2.5119, 'grad_norm': 1.536733865737915, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 1.8856, 'grad_norm': 1.3685024976730347, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 2.8215484619140625, 'eval_runtime': 0.1313, 'eval_samples_per_second': 15.234, 'eval_steps_per_second': 7.617, 'epoch': 2.0}\n",
+      "{'loss': 2.0365, 'grad_norm': 1.2905688285827637, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.6189, 'grad_norm': 1.1336886882781982, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 2.816497802734375, 'eval_runtime': 0.143, 'eval_samples_per_second': 13.982, 'eval_steps_per_second': 6.991, 'epoch': 3.0}\n",
+      "{'train_runtime': 1.9967, 'train_samples_per_second': 3.005, 'train_steps_per_second': 3.005, 'train_loss': 2.2438440124193826, 'epoch': 3.0}\n",
+      "fake ai hoax paper made up by gpt-4\n",
+      "{'before': [7.704880714416504, 6.162439346313477, 2.8038063049316406, 6.360780715942383, 4.05305290222168, 16.83307456970215], 'after': [6.772946834564209, 5.394711494445801, 2.9839484691619873, 6.627538681030273, 3.9177215099334717, 16.98150634765625]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.8982, 'grad_norm': 1.66070556640625, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 4.0169, 'grad_norm': 1.7133033275604248, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 4.369366645812988, 'eval_runtime': 0.1559, 'eval_samples_per_second': 12.828, 'eval_steps_per_second': 6.414, 'epoch': 1.0}\n",
+      "{'loss': 3.5287, 'grad_norm': 1.2428947687149048, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 3.3378, 'grad_norm': 1.4541748762130737, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 4.356468677520752, 'eval_runtime': 0.158, 'eval_samples_per_second': 12.661, 'eval_steps_per_second': 6.33, 'epoch': 2.0}\n",
+      "{'loss': 3.0659, 'grad_norm': 0.93393474817276, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 3.0109, 'grad_norm': 1.3038973808288574, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 4.354191780090332, 'eval_runtime': 0.1594, 'eval_samples_per_second': 12.548, 'eval_steps_per_second': 6.274, 'epoch': 3.0}\n",
+      "{'train_runtime': 2.2731, 'train_samples_per_second': 2.64, 'train_steps_per_second': 2.64, 'train_loss': 3.476408918698629, 'epoch': 3.0}\n",
+      "Hardware Hedging Against Scaling Regime Shifts (self.mlscaling)\n",
+      "{'before': [10.767059326171875, 3.284585475921631, 2.225442886352539, 5.182804107666016], 'after': [10.917644500732422, 3.4139397144317627, 2.3915717601776123, 5.278247833251953]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 2.0168, 'grad_norm': 1.5045863389968872, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 1.9515, 'grad_norm': 1.6995381116867065, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 2.9758172035217285, 'eval_runtime': 0.1776, 'eval_samples_per_second': 11.263, 'eval_steps_per_second': 5.631, 'epoch': 1.0}\n",
+      "{'loss': 1.7056, 'grad_norm': 1.089555263519287, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 1.4582, 'grad_norm': 1.0138260126113892, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 2.888021469116211, 'eval_runtime': 0.1766, 'eval_samples_per_second': 11.327, 'eval_steps_per_second': 5.664, 'epoch': 2.0}\n",
+      "{'loss': 1.3538, 'grad_norm': 1.3632088899612427, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.2299, 'grad_norm': 0.939950168132782, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 2.8587775230407715, 'eval_runtime': 0.178, 'eval_samples_per_second': 11.236, 'eval_steps_per_second': 5.618, 'epoch': 3.0}\n",
+      "{'train_runtime': 2.458, 'train_samples_per_second': 2.441, 'train_steps_per_second': 2.441, 'train_loss': 1.619314471880595, 'epoch': 3.0}\n",
+      "How to Focus\n",
+      "{'before': [11.175151824951172, 3.43853759765625, 8.342260360717773, 8.468477249145508], 'after': [11.017436981201172, 3.291187047958374, 7.970725059509277, 8.81446647644043]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 2.9858, 'grad_norm': 2.995089054107666, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.6439, 'grad_norm': 1.58451247215271, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.429967164993286, 'eval_runtime': 0.1566, 'eval_samples_per_second': 12.768, 'eval_steps_per_second': 6.384, 'epoch': 1.0}\n",
+      "{'loss': 2.4293, 'grad_norm': 1.9177875518798828, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.992, 'grad_norm': 1.236825704574585, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.331143379211426, 'eval_runtime': 0.1591, 'eval_samples_per_second': 12.574, 'eval_steps_per_second': 6.287, 'epoch': 2.0}\n",
+      "{'loss': 1.8766, 'grad_norm': 1.3951705694198608, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.6741, 'grad_norm': 1.2922844886779785, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.304795742034912, 'eval_runtime': 0.1609, 'eval_samples_per_second': 12.431, 'eval_steps_per_second': 6.216, 'epoch': 3.0}\n",
+      "{'train_runtime': 2.2869, 'train_samples_per_second': 2.624, 'train_steps_per_second': 2.624, 'train_loss': 2.7669491370519004, 'epoch': 3.0}\n",
+      "Deontic Explorations In \"Paying To Talk To Slaves\"\n",
+      "{'before': [7.481788158416748, 3.8988535404205322, 4.159244537353516, 13.049931526184082], 'after': [4.832216262817383, 3.738274574279785, 4.002066612243652, 13.358238220214844]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 2.8747, 'grad_norm': 2.096508502960205, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.5899, 'grad_norm': 2.015336751937866, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.5862479209899902, 'eval_runtime': 0.1495, 'eval_samples_per_second': 13.375, 'eval_steps_per_second': 6.687, 'epoch': 1.0}\n",
+      "{'loss': 2.4316, 'grad_norm': 1.262579321861267, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.7612, 'grad_norm': 1.335434079170227, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.575761556625366, 'eval_runtime': 0.1538, 'eval_samples_per_second': 13.007, 'eval_steps_per_second': 6.503, 'epoch': 2.0}\n",
+      "{'loss': 1.8673, 'grad_norm': 1.1828184127807617, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.3797, 'grad_norm': 1.230667233467102, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.5691187381744385, 'eval_runtime': 0.1585, 'eval_samples_per_second': 12.622, 'eval_steps_per_second': 6.311, 'epoch': 3.0}\n",
+      "{'train_runtime': 2.656, 'train_samples_per_second': 2.259, 'train_steps_per_second': 2.259, 'train_loss': 2.650757590929667, 'epoch': 3.0}\n",
+      "By default, capital will matter more than ever after AGI\n",
+      "{'before': [12.33033275604248, 10.804149627685547, 5.264718532562256, 5.00636625289917, 2.185818910598755], 'after': [13.265295028686523, 10.46897029876709, 5.712345600128174, 4.91886043548584, 2.082519292831421]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.7673, 'grad_norm': 2.271059036254883, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 2.6765, 'grad_norm': 1.7324353456497192, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.623122215270996, 'eval_runtime': 0.1843, 'eval_samples_per_second': 10.85, 'eval_steps_per_second': 5.425, 'epoch': 1.0}\n",
+      "{'loss': 3.4631, 'grad_norm': 2.2982003688812256, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.3217, 'grad_norm': 1.8474977016448975, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.6200013160705566, 'eval_runtime': 0.186, 'eval_samples_per_second': 10.755, 'eval_steps_per_second': 5.378, 'epoch': 2.0}\n",
+      "{'loss': 2.9655, 'grad_norm': 1.1774643659591675, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.1688, 'grad_norm': 0.7183546423912048, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.624293565750122, 'eval_runtime': 0.1877, 'eval_samples_per_second': 10.657, 'eval_steps_per_second': 5.328, 'epoch': 3.0}\n",
+      "{'train_runtime': 2.6891, 'train_samples_per_second': 2.231, 'train_steps_per_second': 2.231, 'train_loss': 2.893831968307495, 'epoch': 3.0}\n",
+      "The Plan - 2024 Update\n",
+      "{'before': [11.085138320922852, 7.613076686859131, 7.271832466125488, 2.5242795944213867], 'after': [10.461780548095703, 7.329089164733887, 6.638682842254639, 2.3973302841186523]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 2.4617, 'grad_norm': 2.314183235168457, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.6791, 'grad_norm': 4.621912956237793, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 2.6804332733154297, 'eval_runtime': 0.1894, 'eval_samples_per_second': 10.557, 'eval_steps_per_second': 5.279, 'epoch': 1.0}\n",
+      "{'loss': 1.937, 'grad_norm': 1.5637528896331787, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.8931, 'grad_norm': 3.296773672103882, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 2.6785223484039307, 'eval_runtime': 0.1935, 'eval_samples_per_second': 10.335, 'eval_steps_per_second': 5.167, 'epoch': 2.0}\n",
+      "{'loss': 1.4123, 'grad_norm': 1.2823187112808228, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.5411, 'grad_norm': 1.8647823333740234, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 2.6879079341888428, 'eval_runtime': 0.1877, 'eval_samples_per_second': 10.653, 'eval_steps_per_second': 5.327, 'epoch': 3.0}\n",
+      "{'train_runtime': 2.7828, 'train_samples_per_second': 2.156, 'train_steps_per_second': 2.156, 'train_loss': 2.4873830676078796, 'epoch': 3.0}\n",
+      "Amazon-backed Anthropic debuts AI agents that can do complex tasks, racing against OpenAI, Microsoft and Google\n",
+      "{'before': [7.059969425201416, 7.866158485412598, 4.972960948944092, 1.4543977975845337, 11.503323554992676], 'after': [4.935206890106201, 6.333395004272461, 4.868557929992676, 1.1692737340927124, 11.316805839538574]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.4176, 'grad_norm': 2.2762768268585205, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.2375, 'grad_norm': 2.3001043796539307, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.4287538528442383, 'eval_runtime': 0.1674, 'eval_samples_per_second': 11.949, 'eval_steps_per_second': 5.975, 'epoch': 1.0}\n",
+      "{'loss': 2.8886, 'grad_norm': 1.398822546005249, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.4307, 'grad_norm': 1.3295927047729492, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.4019901752471924, 'eval_runtime': 0.169, 'eval_samples_per_second': 11.832, 'eval_steps_per_second': 5.916, 'epoch': 2.0}\n",
+      "{'loss': 2.3348, 'grad_norm': 1.320399284362793, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.0626, 'grad_norm': 1.3593965768814087, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.4008889198303223, 'eval_runtime': 0.1679, 'eval_samples_per_second': 11.912, 'eval_steps_per_second': 5.956, 'epoch': 3.0}\n",
+      "{'train_runtime': 2.8613, 'train_samples_per_second': 2.097, 'train_steps_per_second': 2.097, 'train_loss': 2.7286322514216104, 'epoch': 3.0}\n",
+      "OpenAI Email Archives from Musk v. Altman\n",
+      "{'before': [9.35188102722168, 3.8830721378326416, 2.591801881790161, 1.859856367111206, 10.639219284057617, 6.248669624328613], 'after': [7.7498250007629395, 3.6932098865509033, 2.136195659637451, 1.9319171905517578, 10.877782821655273, 6.625941276550293]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.5776, 'grad_norm': 4.010377883911133, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.6589, 'grad_norm': 3.6633472442626953, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.2440617084503174, 'eval_runtime': 0.1115, 'eval_samples_per_second': 17.937, 'eval_steps_per_second': 8.968, 'epoch': 1.0}\n",
+      "{'loss': 2.9709, 'grad_norm': 2.407945394515991, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.4008, 'grad_norm': 2.5443220138549805, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.1872379779815674, 'eval_runtime': 0.1116, 'eval_samples_per_second': 17.928, 'eval_steps_per_second': 8.964, 'epoch': 2.0}\n",
+      "{'loss': 2.2253, 'grad_norm': 1.7101964950561523, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.8499, 'grad_norm': 1.7173982858657837, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.1816787719726562, 'eval_runtime': 0.1135, 'eval_samples_per_second': 17.622, 'eval_steps_per_second': 8.811, 'epoch': 3.0}\n",
+      "{'train_runtime': 2.6239, 'train_samples_per_second': 2.287, 'train_steps_per_second': 2.287, 'train_loss': 2.7805453737576804, 'epoch': 3.0}\n",
+      "President Trump Announces Morgan Ortagus as Deputy Special Presidential Envoy for Middle East Peace\n",
+      "{'before': [9.416387557983398, 10.741005897521973, 2.1393628120422363, 13.070490837097168, 1.8174492120742798], 'after': [9.39091968536377, 10.194953918457031, 2.2045726776123047, 13.611124992370605, 1.771489143371582]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 2.8964, 'grad_norm': 2.3559041023254395, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 2.8855, 'grad_norm': 2.407257556915283, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.4159514904022217, 'eval_runtime': 0.1467, 'eval_samples_per_second': 13.638, 'eval_steps_per_second': 6.819, 'epoch': 1.0}\n",
+      "{'loss': 2.2417, 'grad_norm': 1.521421194076538, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.004, 'grad_norm': 1.6388133764266968, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.386564254760742, 'eval_runtime': 0.1444, 'eval_samples_per_second': 13.846, 'eval_steps_per_second': 6.923, 'epoch': 2.0}\n",
+      "{'loss': 1.5761, 'grad_norm': 1.3388304710388184, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.6097, 'grad_norm': 1.3109463453292847, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.395535469055176, 'eval_runtime': 0.1498, 'eval_samples_per_second': 13.355, 'eval_steps_per_second': 6.677, 'epoch': 3.0}\n",
+      "{'train_runtime': 2.7717, 'train_samples_per_second': 2.165, 'train_steps_per_second': 2.165, 'train_loss': 2.202237923940023, 'epoch': 3.0}\n",
+      "CDC Report on Missouri H5N1 Serology Testing\n",
+      "{'before': [9.302383422851562, 2.164090156555176, 2.530121326446533, 8.904608726501465], 'after': [7.715032577514648, 2.296508312225342, 2.2597248554229736, 10.417621612548828]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 4.6645, 'grad_norm': 2.517502546310425, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.5351, 'grad_norm': 1.9744876623153687, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 4.003051280975342, 'eval_runtime': 0.1857, 'eval_samples_per_second': 10.768, 'eval_steps_per_second': 5.384, 'epoch': 1.0}\n",
+      "{'loss': 4.0242, 'grad_norm': 1.5720769166946411, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.7426, 'grad_norm': 1.93308687210083, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.8951995372772217, 'eval_runtime': 0.1857, 'eval_samples_per_second': 10.772, 'eval_steps_per_second': 5.386, 'epoch': 2.0}\n",
+      "{'loss': 3.2325, 'grad_norm': 1.8237318992614746, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.3779, 'grad_norm': 1.5025017261505127, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.867494821548462, 'eval_runtime': 0.1954, 'eval_samples_per_second': 10.234, 'eval_steps_per_second': 5.117, 'epoch': 3.0}\n",
+      "{'train_runtime': 3.1729, 'train_samples_per_second': 1.891, 'train_steps_per_second': 1.891, 'train_loss': 3.429463267326355, 'epoch': 3.0}\n",
+      "2024 in AI predictions\n",
+      "{'before': [8.183303833007812, 0.7180596590042114, 8.179237365722656, 4.342537879943848, 10.570826530456543], 'after': [5.38284969329834, 0.9092058539390564, 8.8030424118042, 4.543374061584473, 10.412266731262207]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 4.0063, 'grad_norm': 2.618964672088623, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.0173, 'grad_norm': 3.8962526321411133, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 4.496792793273926, 'eval_runtime': 0.1718, 'eval_samples_per_second': 11.638, 'eval_steps_per_second': 5.819, 'epoch': 1.0}\n",
+      "{'loss': 3.2503, 'grad_norm': 2.181051731109619, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.2413, 'grad_norm': 1.5369327068328857, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 4.481138229370117, 'eval_runtime': 0.1743, 'eval_samples_per_second': 11.475, 'eval_steps_per_second': 5.737, 'epoch': 2.0}\n",
+      "{'loss': 2.4798, 'grad_norm': 1.650524377822876, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.8449, 'grad_norm': 1.324534296989441, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 4.495326995849609, 'eval_runtime': 0.1831, 'eval_samples_per_second': 10.923, 'eval_steps_per_second': 5.462, 'epoch': 3.0}\n",
+      "{'train_runtime': 3.2972, 'train_samples_per_second': 1.82, 'train_steps_per_second': 1.82, 'train_loss': 2.806656757990519, 'epoch': 3.0}\n",
+      "Comment on 'Death and the Gorgon'\n",
+      "{'before': [11.654606819152832, 3.051333427429199, 9.083542823791504, 6.945958137512207, 10.134012222290039], 'after': [12.029337882995605, 2.9569177627563477, 9.219869613647461, 7.349301338195801, 10.291199684143066]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.6214, 'grad_norm': 2.26008939743042, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.025, 'grad_norm': 2.695547580718994, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.424791097640991, 'eval_runtime': 0.1737, 'eval_samples_per_second': 11.514, 'eval_steps_per_second': 5.757, 'epoch': 1.0}\n",
+      "{'loss': 2.9282, 'grad_norm': 1.7743608951568604, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.0993, 'grad_norm': 1.5769550800323486, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.3480515480041504, 'eval_runtime': 0.1733, 'eval_samples_per_second': 11.538, 'eval_steps_per_second': 5.769, 'epoch': 2.0}\n",
+      "{'loss': 2.2028, 'grad_norm': 1.4857354164123535, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.6797, 'grad_norm': 1.4504806995391846, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.32250714302063, 'eval_runtime': 0.187, 'eval_samples_per_second': 10.695, 'eval_steps_per_second': 5.348, 'epoch': 3.0}\n",
+      "{'train_runtime': 3.2739, 'train_samples_per_second': 1.833, 'train_steps_per_second': 1.833, 'train_loss': 2.5927275021870932, 'epoch': 3.0}\n",
+      "debating buying NVDA in 2019\n",
+      "{'before': [7.8792829513549805, 5.101133346557617, 0.25373393297195435, 7.965005397796631], 'after': [4.433870792388916, 4.562448501586914, 0.215262308716774, 8.595725059509277]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.4299, 'grad_norm': 2.6792428493499756, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.3305, 'grad_norm': 2.825462818145752, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.3379433155059814, 'eval_runtime': 0.1741, 'eval_samples_per_second': 11.485, 'eval_steps_per_second': 5.742, 'epoch': 1.0}\n",
+      "{'loss': 2.7355, 'grad_norm': 1.6066468954086304, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.3417, 'grad_norm': 1.7739102840423584, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.325791597366333, 'eval_runtime': 0.1665, 'eval_samples_per_second': 12.009, 'eval_steps_per_second': 6.005, 'epoch': 2.0}\n",
+      "{'loss': 1.9831, 'grad_norm': 1.7341994047164917, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.8483, 'grad_norm': 1.8331834077835083, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.3339157104492188, 'eval_runtime': 0.1744, 'eval_samples_per_second': 11.469, 'eval_steps_per_second': 5.735, 'epoch': 3.0}\n",
+      "{'train_runtime': 3.3289, 'train_samples_per_second': 1.802, 'train_steps_per_second': 1.802, 'train_loss': 2.611490547657013, 'epoch': 3.0}\n",
+      "Human study on AI spear phishing campaigns\n",
+      "{'before': [8.617542266845703, 5.865469455718994, 5.855831146240234, 6.869799613952637, 0.7429558634757996, 5.193291664123535], 'after': [4.326836585998535, 5.962702751159668, 6.1263532638549805, 7.080767631530762, 0.37877461314201355, 5.513457298278809]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.0659, 'grad_norm': 2.3972017765045166, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.3097, 'grad_norm': 3.591853618621826, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.3992388248443604, 'eval_runtime': 0.1902, 'eval_samples_per_second': 10.513, 'eval_steps_per_second': 5.256, 'epoch': 1.0}\n",
+      "{'loss': 2.3208, 'grad_norm': 1.6523462533950806, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.4739, 'grad_norm': 3.74638032913208, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.375133514404297, 'eval_runtime': 0.1871, 'eval_samples_per_second': 10.691, 'eval_steps_per_second': 5.346, 'epoch': 2.0}\n",
+      "{'loss': 1.5714, 'grad_norm': 1.4942113161087036, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.0755, 'grad_norm': 2.089600086212158, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.3879997730255127, 'eval_runtime': 0.1873, 'eval_samples_per_second': 10.679, 'eval_steps_per_second': 5.339, 'epoch': 3.0}\n",
+      "{'train_runtime': 3.3319, 'train_samples_per_second': 1.801, 'train_steps_per_second': 1.801, 'train_loss': 2.469515860080719, 'epoch': 3.0}\n",
+      "My AGI safety research—2024 review, ’25 plans\n",
+      "{'before': [10.924260139465332, 4.072440147399902, 2.9389638900756836, 0.10930612683296204, 5.323551177978516], 'after': [10.128201484680176, 2.6714420318603516, 2.3213531970977783, 0.08397671580314636, 5.336966514587402]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.5354, 'grad_norm': 2.6700191497802734, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.1829, 'grad_norm': 3.94592547416687, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 4.003451824188232, 'eval_runtime': 0.1551, 'eval_samples_per_second': 12.895, 'eval_steps_per_second': 6.447, 'epoch': 1.0}\n",
+      "{'loss': 2.7582, 'grad_norm': 1.6923402547836304, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.2478, 'grad_norm': 3.04793643951416, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 4.013311862945557, 'eval_runtime': 0.1558, 'eval_samples_per_second': 12.838, 'eval_steps_per_second': 6.419, 'epoch': 2.0}\n",
+      "{'loss': 1.8756, 'grad_norm': 1.6792405843734741, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.7245, 'grad_norm': 2.353637933731079, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 4.035435199737549, 'eval_runtime': 0.1585, 'eval_samples_per_second': 12.617, 'eval_steps_per_second': 6.309, 'epoch': 3.0}\n",
+      "{'train_runtime': 3.2121, 'train_samples_per_second': 1.868, 'train_steps_per_second': 1.868, 'train_loss': 2.5540746450424194, 'epoch': 3.0}\n",
+      "Parkinson's Law and the Ideology of Statistics\n",
+      "{'before': [11.059673309326172, 0.410934716463089, 4.9378437995910645, 9.082354545593262], 'after': [11.269186019897461, 0.23422664403915405, 5.095401763916016, 9.99195671081543]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.339, 'grad_norm': 2.887451171875, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.7994, 'grad_norm': 3.2563376426696777, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.6911461353302, 'eval_runtime': 0.1747, 'eval_samples_per_second': 11.449, 'eval_steps_per_second': 5.725, 'epoch': 1.0}\n",
+      "{'loss': 2.4807, 'grad_norm': 2.2783405780792236, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.7528, 'grad_norm': 3.188812494277954, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.6750943660736084, 'eval_runtime': 0.1684, 'eval_samples_per_second': 11.876, 'eval_steps_per_second': 5.938, 'epoch': 2.0}\n",
+      "{'loss': 1.591, 'grad_norm': 2.149153470993042, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.279, 'grad_norm': 1.790748953819275, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.6802124977111816, 'eval_runtime': 0.1716, 'eval_samples_per_second': 11.652, 'eval_steps_per_second': 5.826, 'epoch': 3.0}\n",
+      "{'train_runtime': 3.6872, 'train_samples_per_second': 1.627, 'train_steps_per_second': 1.627, 'train_loss': 2.7069696386655173, 'epoch': 3.0}\n",
+      "Preference Inversion\n",
+      "{'before': [11.257996559143066, 2.4081733226776123, 7.490055084228516, 6.412927627563477], 'after': [12.913603782653809, 2.2612903118133545, 7.073019981384277, 8.057035446166992]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.5488, 'grad_norm': 3.1160054206848145, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 4.0606, 'grad_norm': 3.991938591003418, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.7835068702697754, 'eval_runtime': 0.2216, 'eval_samples_per_second': 9.024, 'eval_steps_per_second': 4.512, 'epoch': 1.0}\n",
+      "{'loss': 2.808, 'grad_norm': 2.174905300140381, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 3.2191, 'grad_norm': 7.194904804229736, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.7082245349884033, 'eval_runtime': 0.2222, 'eval_samples_per_second': 8.999, 'eval_steps_per_second': 4.5, 'epoch': 2.0}\n",
+      "{'loss': 2.043, 'grad_norm': 1.584335207939148, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.8653, 'grad_norm': 2.442568778991699, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.646120548248291, 'eval_runtime': 0.2322, 'eval_samples_per_second': 8.613, 'eval_steps_per_second': 4.307, 'epoch': 3.0}\n",
+      "{'train_runtime': 4.3763, 'train_samples_per_second': 1.371, 'train_steps_per_second': 1.371, 'train_loss': 3.09080179532369, 'epoch': 3.0}\n",
+      "Review: Planecrash\n",
+      "{'before': [8.528902053833008, 13.35799789428711, 8.998847961425781, 1.961258888244629, 0.19278351962566376], 'after': [9.244955062866211, 12.997112274169922, 4.955875396728516, 0.008964293636381626, 0.050971198827028275]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.7415, 'grad_norm': 2.612143039703369, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 2.835, 'grad_norm': 3.254910469055176, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.560338258743286, 'eval_runtime': 0.1714, 'eval_samples_per_second': 11.669, 'eval_steps_per_second': 5.835, 'epoch': 1.0}\n",
+      "{'loss': 2.8215, 'grad_norm': 2.334237575531006, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 1.6291, 'grad_norm': 1.8375204801559448, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.4723968505859375, 'eval_runtime': 0.171, 'eval_samples_per_second': 11.698, 'eval_steps_per_second': 5.849, 'epoch': 2.0}\n",
+      "{'loss': 1.9053, 'grad_norm': 2.1019699573516846, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.1471, 'grad_norm': 1.539451241493225, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.4505653381347656, 'eval_runtime': 0.176, 'eval_samples_per_second': 11.361, 'eval_steps_per_second': 5.681, 'epoch': 3.0}\n",
+      "{'train_runtime': 3.9229, 'train_samples_per_second': 1.529, 'train_steps_per_second': 1.529, 'train_loss': 2.346584419409434, 'epoch': 3.0}\n",
+      "The Field of AI Alignment: A Postmortem, and What To Do About It\n",
+      "{'before': [10.715858459472656, 2.8538293838500977, 9.619916915893555, 2.8236703872680664, 1.3537986278533936, 9.357948303222656], 'after': [17.13212776184082, 2.7473714351654053, 9.264034271240234, 3.5286686420440674, 0.48276376724243164, 9.244832992553711]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.3063, 'grad_norm': 3.2994930744171143, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.6343, 'grad_norm': 3.1291298866271973, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.973965644836426, 'eval_runtime': 0.1735, 'eval_samples_per_second': 11.525, 'eval_steps_per_second': 5.763, 'epoch': 1.0}\n",
+      "{'loss': 2.4674, 'grad_norm': 2.477537155151367, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.3556, 'grad_norm': 2.0268757343292236, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.917025566101074, 'eval_runtime': 0.1726, 'eval_samples_per_second': 11.589, 'eval_steps_per_second': 5.795, 'epoch': 2.0}\n",
+      "{'loss': 1.5868, 'grad_norm': 1.8040292263031006, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.7624, 'grad_norm': 1.7323975563049316, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.9284799098968506, 'eval_runtime': 0.1773, 'eval_samples_per_second': 11.28, 'eval_steps_per_second': 5.64, 'epoch': 3.0}\n",
+      "{'train_runtime': 3.5944, 'train_samples_per_second': 1.669, 'train_steps_per_second': 1.669, 'train_loss': 2.518795907497406, 'epoch': 3.0}\n",
+      "The Intelligence Curse\n",
+      "{'before': [11.011114120483398, 0.6101000308990479, 8.701272964477539, 2.934598207473755, 10.553529739379883], 'after': [11.304214477539062, 0.4481358230113983, 7.420818328857422, 2.2388195991516113, 10.681926727294922]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 2.983, 'grad_norm': 3.2141897678375244, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.3046, 'grad_norm': 3.5254054069519043, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 2.876291275024414, 'eval_runtime': 0.1805, 'eval_samples_per_second': 11.079, 'eval_steps_per_second': 5.54, 'epoch': 1.0}\n",
+      "{'loss': 2.1398, 'grad_norm': 2.1053242683410645, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.0996, 'grad_norm': 1.6926547288894653, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 2.890993595123291, 'eval_runtime': 0.1859, 'eval_samples_per_second': 10.761, 'eval_steps_per_second': 5.381, 'epoch': 2.0}\n",
+      "{'loss': 1.3094, 'grad_norm': 1.8891417980194092, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.5913, 'grad_norm': 1.575945496559143, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 2.9214653968811035, 'eval_runtime': 0.1873, 'eval_samples_per_second': 10.677, 'eval_steps_per_second': 5.338, 'epoch': 3.0}\n",
+      "{'train_runtime': 3.8221, 'train_samples_per_second': 1.57, 'train_steps_per_second': 1.57, 'train_loss': 2.2379422187805176, 'epoch': 3.0}\n",
+      "The Laws of Large Numbers\n",
+      "{'before': [11.152969360351562, 12.381094932556152, 6.883550643920898, 5.000158309936523], 'after': [10.98746109008789, 12.829581260681152, 6.23975944519043, 4.902842998504639]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.4465, 'grad_norm': 3.0251810550689697, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 2.4625, 'grad_norm': 3.3443939685821533, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.2333736419677734, 'eval_runtime': 0.1783, 'eval_samples_per_second': 11.22, 'eval_steps_per_second': 5.61, 'epoch': 1.0}\n",
+      "{'loss': 2.6116, 'grad_norm': 2.0492935180664062, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 1.4007, 'grad_norm': 1.8274493217468262, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.223329782485962, 'eval_runtime': 0.1792, 'eval_samples_per_second': 11.162, 'eval_steps_per_second': 5.581, 'epoch': 2.0}\n",
+      "{'loss': 1.7269, 'grad_norm': 1.6054573059082031, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 0.9109, 'grad_norm': 1.4826263189315796, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.2280447483062744, 'eval_runtime': 0.1846, 'eval_samples_per_second': 10.834, 'eval_steps_per_second': 5.417, 'epoch': 3.0}\n",
+      "{'train_runtime': 3.8591, 'train_samples_per_second': 1.555, 'train_steps_per_second': 1.555, 'train_loss': 2.0931734144687653, 'epoch': 3.0}\n",
+      "The subset parity learning problem: much more than you wanted to know\n",
+      "{'before': [6.899591445922852, 0.8442123532295227, 0.49031174182891846, 9.700416564941406], 'after': [5.6730499267578125, 0.8620076775550842, 0.47212398052215576, 10.04554271697998]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 4.1168, 'grad_norm': 3.2236294746398926, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.8383, 'grad_norm': 3.7380311489105225, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 2.892657995223999, 'eval_runtime': 0.1945, 'eval_samples_per_second': 10.282, 'eval_steps_per_second': 5.141, 'epoch': 1.0}\n",
+      "{'loss': 3.2994, 'grad_norm': 2.064343214035034, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.7109, 'grad_norm': 2.8733603954315186, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 2.893580198287964, 'eval_runtime': 0.1934, 'eval_samples_per_second': 10.341, 'eval_steps_per_second': 5.17, 'epoch': 2.0}\n",
+      "{'loss': 2.3681, 'grad_norm': 2.0598092079162598, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 2.1504, 'grad_norm': 2.0436267852783203, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 2.8934695720672607, 'eval_runtime': 0.1953, 'eval_samples_per_second': 10.239, 'eval_steps_per_second': 5.12, 'epoch': 3.0}\n",
+      "{'train_runtime': 4.0349, 'train_samples_per_second': 1.487, 'train_steps_per_second': 1.487, 'train_loss': 3.080639600753784, 'epoch': 3.0}\n",
+      "What’s the short timeline plan?\n",
+      "{'before': [9.168852806091309, 3.9152371883392334, 6.799325942993164, 6.304879188537598], 'after': [4.647651195526123, 4.194304466247559, 7.104080677032471, 6.2407636642456055]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 2.9721, 'grad_norm': 6.88749885559082, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 2.8245, 'grad_norm': 6.573474407196045, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 2.963982105255127, 'eval_runtime': 0.2374, 'eval_samples_per_second': 8.426, 'eval_steps_per_second': 4.213, 'epoch': 1.0}\n",
+      "{'loss': 2.7296, 'grad_norm': 6.738276481628418, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 2.2587, 'grad_norm': 5.604180812835693, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 2.7031450271606445, 'eval_runtime': 0.2389, 'eval_samples_per_second': 8.372, 'eval_steps_per_second': 4.186, 'epoch': 2.0}\n",
+      "{'loss': 2.0789, 'grad_norm': 5.735146522521973, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.7098, 'grad_norm': 3.3593287467956543, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 2.540477991104126, 'eval_runtime': 0.2464, 'eval_samples_per_second': 8.117, 'eval_steps_per_second': 4.059, 'epoch': 3.0}\n",
+      "{'train_runtime': 4.1935, 'train_samples_per_second': 1.431, 'train_steps_per_second': 1.431, 'train_loss': 2.4289310773213706, 'epoch': 3.0}\n",
+      "Lorem ipsum\n",
+      "{'before': [13.19483470916748, 4.822676658630371, 4.694229602813721, 0.15062260627746582, 0.009142794646322727], 'after': [12.946345329284668, 4.517403602600098, 5.635333061218262, 0.7315716743469238, 0.001257344032637775]}\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/training_args.py:1575: FutureWarning: `evaluation_strategy` is deprecated and will be removed in version 4.46 of 🤗 Transformers. Use `eval_strategy` instead\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'loss': 3.3232, 'grad_norm': 2.512242555618286, 'learning_rate': 0.001, 'epoch': 0.5}\n",
+      "{'loss': 3.0244, 'grad_norm': 2.4470341205596924, 'learning_rate': 0.0008, 'epoch': 1.0}\n",
+      "{'eval_loss': 3.506387233734131, 'eval_runtime': 0.1841, 'eval_samples_per_second': 10.862, 'eval_steps_per_second': 5.431, 'epoch': 1.0}\n",
+      "{'loss': 2.4132, 'grad_norm': 1.939867615699768, 'learning_rate': 0.0006, 'epoch': 1.5}\n",
+      "{'loss': 1.8838, 'grad_norm': 2.473356246948242, 'learning_rate': 0.0004, 'epoch': 2.0}\n",
+      "{'eval_loss': 3.5382893085479736, 'eval_runtime': 0.1772, 'eval_samples_per_second': 11.287, 'eval_steps_per_second': 5.644, 'epoch': 2.0}\n",
+      "{'loss': 1.4453, 'grad_norm': 1.9782445430755615, 'learning_rate': 0.0002, 'epoch': 2.5}\n",
+      "{'loss': 1.3757, 'grad_norm': 1.529227375984192, 'learning_rate': 0.0, 'epoch': 3.0}\n",
+      "{'eval_loss': 3.5465643405914307, 'eval_runtime': 0.1858, 'eval_samples_per_second': 10.765, 'eval_steps_per_second': 5.382, 'epoch': 3.0}\n",
+      "{'train_runtime': 3.7878, 'train_samples_per_second': 1.584, 'train_steps_per_second': 1.584, 'train_loss': 2.2442721724510193, 'epoch': 3.0}\n",
+      "politics is the mind-killer\n",
+      "{'before': [11.374530792236328, 3.393611192703247, 13.047080993652344, 3.0926449298858643, 6.4303483963012695, 8.23534870147705], 'after': [11.683854103088379, 3.5927183628082275, 13.514053344726562, 3.821317672729492, 7.558355808258057, 7.392338275909424]}\n"
+     ]
+    }
+   ],
+   "source": [
+    "data = []\n",
+    "for i in tqdm(range(len(df))):\n",
+    "    sample = df.iloc[i]\n",
+    "    r = learn_sample(sample)\n",
+    "    print(sample['title'])\n",
+    "    print(dict(before=r['before'], after=r['after']))\n",
+    "    data.append(dict(**r, **sample))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
       "text/plain": [
        "<Figure size 800x550 with 2 Axes>"
       ]
@@ -972,32 +1283,11 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 19,
+   "execution_count": 21,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "<Axes: xlabel='epoch'>"
-      ]
-     },
-     "execution_count": 19,
-     "metadata": {},
-     "output_type": "execute_result"
-    },
-    {
-     "data": {
-      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAqYAAAHmCAYAAAChy0xuAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8g+/7EAAAACXBIWXMAAA9hAAAPYQGoP6dpAABM70lEQVR4nO3deXydZYH+/+ss2fd9T9qmSZq2SZvu0NLSIpuAIoogAo6izDC4IQ4MfH8z4jaM8v06bigKVhlBWWTHBa2UpbSlNE2XpEuWpmn2fV/OyTnn+f1xkkhFoGlz8jxJPu/Xi5ftyWlypbdPe/V+7vt+bIZhGAIAAABMZjc7AAAAACBRTAEAAGARFFMAAABYAsUUAAAAlkAxBQAAgCVQTAEAAGAJFFMAAABYAsUUAAAAluA0O8DZam/vf8+P2+02xcdHqKtrUD4fzxIwA2NgPsbAfIyBNTAO5mMMzGfGGCQlRZ3W+2b9jKndbpPNZpPdbjM7ypzFGJiPMTAfY2ANjIP5GAPzWXkMZn0xBQAAwMxAMQUAAIAlUEwBAABgCRRTAAAAWALFFAAAAJZAMQUAAIAlUEwBAABgCRRTAAAAWALFFAAAAJZg+iNJR0dHdeedd6qlpUVhYWG67777FB8fb3YsAAAATDPTZ0z/8Ic/KCUlRb/5zW902WWX6cEHHzQ7EgAAAExg+ozphz/8YV122WWSpJaWFsXExJicCAAAAGaYtmL6+OOP69e//vUpr/3iF79QSkqKnE6nbr75Zh06dEi//OUvpysSAAAALGTaiuk111yja6655l0//vOf/1x1dXW6+eab9dJLL01XLAAAAFiE6WtMH3/8cT3yyCOSpPDwcNntpkcCAACACc64Bbrdbl1++eV68803J15zuVy6++67tWrVKm3YsEFbt259389z6aWXaseOHbr++uv1pS99Sd/85jfPNBIAAABmsDO6le9yuXT77berqqrqlNe/+93vqry8XA8//LCampp05513Kj09XZdccsm7fq7o6Gg98MADZxJDkmS322S329714w6H/ZT/na3+uLtONY29mpcWrbzMGM1Pi1ZwkMPsWJLmzhhYGWNgPsbAGhgH8zEG5rPyGEy6mFZXV+v222+XYRinvD40NKQnn3xSDz74oJYsWaIlS5aoqqpKjz766HsW07MVHx8hm+3di+m46OiwgGUwW1V9t367zf+PhD1H2iRJDrtNuZkxWjQvXoVj/yXEmPt7MJvHYKZgDMzHGFgD42A+xsB8VhyDSRfTPXv2aO3atbrtttu0fPnyidePHj0qj8ejkpKSiddWrlypBx54QD6fL2BrR7u6Bt93xjQ6Okx9fcPyen0ByWC2Z7dXS5LsNpsMGTIMyeszVHmyR5Une/T8a8clSQnRocrLilFeZqzyMmOUlRIpxzSs6Z0LY2B1jIH5GANrYBzMxxiYz4wxiIuLOK33TbqYXnfddf/w9fb2dsXFxSk4OHjitcTERLlcLvX09ATsaU4+nyGfz3jf93m9Pnk8s+8CGBwZ1e6KFknShuJUXXtBnmqb+lTd2Kvqxj7VNPZqyOWRJHX2jaizYkS7K1olScFBdi1Ii9bCzBgtzIjRgvQYRYYFBSzrbB2DmYQxMB9jYA2Mg/kYA/NZcQym7Lio4eHhU0qppImfu93uqfoy+Ds7D7XIPfZ/qs0lmQoNdk7cupckn2GouXNINY29qm7oVVVjr1q7hiRJ7lGfjp7s0dGTPROfLy0hXAsz/EU1LytWKXFhp7VUAgAA4GxNWTENCQl5RwEd/3loaOhUfRm8jWEY2l7WKElakB6tnNSod7zHbrMpIzFCGYkR2rgsXZLUP+RWTeP4rGqvapv7NDpWbps7h9TcOaTXDzZLkqLDgyZu/edlxSp7mm7/AwCAuWfKimlKSoq6u7vl8XjkdPo/bXt7u0JDQxUdHT1VXwZvc7SuWy1js5+bSzJO+9dFhQdreV6iluclSpI8Xp/q2wZU3dA7UVa7+12SpL6hUZVWtqu0sl2SFBLk0IJ0/87//KxYLUiPVmiw6U+2BQAAs8CUNYrCwkI5nU7t379fq1atkiSVlpaqqKiIQ/MDZHy2NCLUqdWLks/48zgdds1Pi9b8tGhduDpLktTZO6Kqhh5VNfSqsqFHje2DkiTXqFdH6rp1pK5bkn9GNic18m+zqpmxio4IftevBQAA8G6mrJiGhYXpyiuv1D333KP/+q//Ultbm7Zu3ap77713qr4E3qZnwKWyqg5J0vqitCk/szQhJlQJMalatyRVkn+TVfVYSa1q6NWJ5j55vIZ8hqHa5n7VNvfrz2/VS5JS4sP9M6qZscrLilF64untxAMAAHPblN6Dveuuu3TPPffoU5/6lCIjI/WFL3xBF1100VR+CYx57UCTvGOnEUzmNv6ZiggN0rKFiVq20H/7f9TjVW1z/8SsalVDr4bHdv+3dg2ptWtIO8bWqcZEBmvpgkQtSItSXkaM0pMiZGdDFQAA+Ds24+9Pyp9h2tv73/PjTqddcXER6u4etNyRCGfK6/Ppjp/uUne/S4vnxemr15a8/y8KMJ/PUGPHoKoaelRZ7y+r4+tU/15EqFP5WbEqyIpVQXacspIj3/MsWpy92XgdzDSMgTUwDuZjDMxnxhgkJb1zg/Y/wq6VGehgdedE6ZuO2dLTYbfblJUcqazkSG1ZkSnDMNTZN6Kqev9mqqrGXjW0DUiSBkc8KqvqmFiKEBbiUF6mv6jmZ8cqJyVKTgs+Jg0AAAQWxXQGGt/0FBsZPLGz3mpsNpsSY8KUGBOm85anKy4uQnUN3Tpc26VjJ3t0rL5bDWMbqoZdXh2s6dTBmk5J/p3/CzOilZ8dp4KsWM1Pi1aQk6IKAMBsRzGdYdq6h1Re2yVJ2rgsfUadKRodEaxVi5K1auwEgYHhUVXV9+hYfY+OnezRybZ+GYZ/53/FiW5VnPDv/A9y2pWbHu2//Z8dp9z06Cnf7AUAAMxHMZ1hXtnfJMl/TNOm5da4jX+mIsOCVJKfpJL8JEnS0IhHVQ3+olpZ36MTzf3yGYZGPW97QtUbJ+R02LQgPUaFOXFalB2rBekxzKgCADALUExnkFGPd2Kn+/K8RMVFhZicaGqFhzpP2fk/4vaourF37NZ/j2qb+uT1GfJ4DVWOldfnJAU77VqYOV5U4zQvLWpGzSQDAAA/iukMsvdouwaGRyVZZ9NTIIUGO7V0foKWzk+QJLlHvapp6tOxk906WtetmrGi6vb4dPhEtw6P3foPDXYoPytWi7LjVJjDrn8AAGYKiukMMr7pKTkuTIXz4kxOM/2CgxwqzPGXTZ0nudxeVTf26khdt46e7FZtc58MQxpxn7qZavx4qsKcOC3KiVNGYoRsnKMKAIDlUExniPq2AVU39kqSzl+ewQH1kkKCHVoyP15L5sdLkoZdHlXW90wU1frWARl65/FU0eFBKhibTV2UE6eUuDCKKgAAFkAxnSHGZ0udDrs2FKeZnMaawkJOXaM6MDyqYyd7dHSsqDZ2+I+n6hsa1VtH2/TW0TZJUlxUiBbPi9OSefEqnBevmIhg074HAADmMorpDDDs8mhXRYskaU1hsiLDgkxONDNEhgVpZUGSVhb4d/33Drp17GS3f0a1rlut3cOSpO5+l9441KI3Dvl/jzOTIrR4XrwWz4tXQVasQoI5mgoAgOlAMZ0Bdle0yOX2Spobm54CJSYiWGsKU7SmMEWS1NU3oiN1/qJacaJLvQNuSVJD+6Aa2gf157fq5bDbtDAjRovnxWnx/HjNS2XHPwAAgUIxtTjDMPTy2G387ORILUiPNjnR7BEfHar1RWlaX5QmwzDU1Dmkwye6dLi2S0fre+Rye+X1Gf4HANT36JnXaxUW4lRhTpy/qM6LZ30qAABTiGJqcVUNvWoce3Tn+SsyKEEBYrPZlJEYoYzECF24Kkser0+1zX2qqO3S4bpuHW/sk88wNOzyaF9lu/ZVtkuSEqJDVDgv3l9Uc+IVzfpUAADOGMXU4l4Zmy0NDXZo3eIUk9PMHU6HXXmZscrLjNWV5/nX+R472aPDJ7pUcaJLzZ1DkqTOPpd2HGyeePBBVnKkls6P19IFCcrLjJHTwW1/AABOF8XUwvqG3Np7zL9z/NylqQoNZrjMEhbi1PK8RC3P8+/47+53+W/7n+jS4RPd6h30r0+tbxtQfduA/vjmSYWMnbu6dEG8ls6PV3JcuJnfAgAAlkfTsbAdB5vl8RqS2PRkNXFRIaeuT+0YVMWJblXUdunYyW65PT65Rr3aX92h/dX+81OT48ImZlMXZcfyDw0AAP4OfzNalM8wJm7j52fGKCMp0uREeDc2m00ZSZHKSIrURauzNOrxqrK+V4eOd6qitmvi/NS27mG93N2ol/c1ymG3KT8rdmw2NUGZSTyNCgAAiqlFlR/vUkfviCT/pifMHEHOU59I1dU3ovLaLpUf71TFiW4Nuzzy+oyJo6qe3F6j2MhgLZkfr6IFCVo8L56zagEAcxLF1KLGZ0ujw4O0Mj/Z5DQ4G/HRodq4LF0bl6XL6/Optqlfh453qry2Syea+2RI6hlwTxzyb5M0Pz1aS8eK6vy0aNntzKYCAGY/iqkFdfaO6ECNf13iecvSFeRkZ/ds4bDbtTAzRgszY/SRjQvUP+TW4RPdKh8rqr2DbhmSjjf16XhTn55/44SiwoNUtCBBxbkJWjo/XuGhzKYCAGYniqkFvXqgUYYh2SRtWpZudhwEUFR4sNYuTtHaxSkyDEP1bQMTt/2rGnrl9RnqHxrVzvIW7Sxvkd1mU15mjIoXJqg4N1HpCeGsTQUAzBoUU4vxeH167YD/TMyi3AQlxoaZnAjTxWazKTslStkpUfrguhwNuzw6fKJbB2o6dKimU72DbvmMvz2J6sntNUqMCVVxrr+kLsqOVXCQw+xvAwCAM0YxtZh9le3qGzsTkyOi5rawEKdWFiRpZUGSfIahk639OljdqQM1nRNrUzt6R/TyPv9O/2CnXYU5cSpemKhluQmKjw41+1sAAGBSKKYWM77pKSE6VEULEkxOA6uw22yalxqteanR+tCG+eobdOvQcX9Jrajt1LDLK7fHpwM1/td+LSkzKULFuYkqzk1QQU6s2d8CAADvi2JqIU0dgzp6skeSdH5JOjux8a6iI4InDvj3eH2qbujVwZpOHajpmHhcakP7oBraB/WH3XWKCAvSqkUpWjIvVkvmxSsshEsfAGA9/O1kIeOzpQ67TecVs+kJp8fpsGtRTpwW5cTp41sWqq1nWIdqOnWgukNHT/bI4/VpcHhUr5Y16NWyBjnsNi3KiVNJXqKWL0zklj8AwDIophbhcnv1RnmLJGllQZKiI4JNToSZKjk2TBeszNQFKzPlcnt1uK5L5ce7dKCmU119I/L6DFXUdqmitkuP/LlSOalRKslLVEleEk+gAgCYimJqEW8eadWwyyOJTU+YOiHBDpXkJWl1YYpiYsJVdqRZe4+2qayqQ43t/kel1rX0q66lX8++XqvEmFAtHyup+Vkxctg5QxcAMH0ophZgGIa27/Pfxs9IjFB+Vqy5gTAr2e02LUiPUXZylK7amKu27iHtr+pQWVWHKht6ZBj+Xf7b9jZo294GRYQ6VZyboJK8JC2Zz7pUAEDg8TeNBdQ296uutV+SdH5JBrdSMS2S48J10ZpsXbQmWwPDozpQ3aH9VR0qr+2Sa9SrwRGPdlW0aldFq5wOmwpz4lWSl6hlCxMVFxVidnwAwCxEMbWA7WUNkqTgILvOWZJqchrMRZFhQRO7/Ec9Xh0+0a2yqg7tr+5Q36BbHq+hQ8c7deh4p/TSMc1Pi9aK/EStKkhWSny42fEBALMExdRkA8Oj2nOkTZK0bnGqwkMZEpgryOnQsoX+mVGfYai2qU9lVR0qq2qfOIqqtrlPtc19eurV48pMitDKgmStKkhSeiKbpwAAZ44WZLKdh5o16vFJYtMTrMdusyk3I0a5GTH62Pm5au0aUllVh/ZVtaumoVeGxs9LrdVzO2qVGh+ulQVJWlWQrOyUSEoqAGBSKKYmMgxD2/c3SZIWpEcrJzXK5ETAe0uJD9cla7N1ydps9Qy4tK+yXaXH2nX0ZLcMQ2rpGtLvd9Xp97vqlBgTOlFS56dHy05JBQC8D4qpiY7Udau1y39rlNlSzDSxkSHasiJTW1Zkqn/IrbKqDpUea9fhE13y+gx19I7opT31emlPveKiQrQiP0mrCpKUlxnLU80AAP8QxdRE28ee9BQR6tTqRckmpwHOXFR4sDYuS9fGZekaGhnV/mp/ST10vEser0/d/S79tbRBfy1tUHR4kEry/TOpBdmxcjo4KxUA4EcxNUl3v0tllR2SpPVFaQoOcpicCJga4aFBOndpms5dmqYRt0cHazq191i7DtV0yjXqVd/QqF7d36RX9zcpItSp5Xn+3f1L5sdTUgFgjqOYmmRnebN8hiGJ2/iYvUKDnVpTmKI1hSlyj3pVXtul0mNt2l/dqWGXR4MjHr1xqEVvHGpRRKhTKwuStKYwRQXZsTx1CgDmIIqpSY7WdUvyb3riHEjMBcFBDq3IT9KK/CR5vD4dPtGt0mP+x6MODI9qcMSj1w4067UDzYoOD9KqRclaU5iihZkxbJwCgDmCYmoCr8+n6sY+SeLxo5iTnA67inMTVJyboBu8Ph2t69abR1q1r7Jdwy7/7f6X9zXq5X2NiosK0ZpCf0mdlxrFEVQAMItRTE1wsnVArlGvJIop4HTYtXRBgpYuSNCNF/tUfrxTbx5p1f7qDrlH/Runxnf3J8eGac1if0nNTIo0OzoAYIpRTE1QWd8jSbJJysuMMTULYCVBTrtK8pNUkp8kl9urAzUd2nOkTQdrOuXx+tTWM6wXd9bpxZ11ykiMmJhJZTkMAMwOFFMTjBfTjKRIRYQGmRsGsKiQYMfExqmhEY/Kqtq150ibKmq75DMMNXYM6pnXa/XM67XKSY3S2sIUrV6UrISYULOjAwDOEMV0mvkMQ1UNvZKk/CxmS4HTER7q1PqiNK0vSlP/kFulle3ac7hVx072yJBU19KvupZ+PbG9WgszYrRuib/QRobxDz8AmEkoptOsuWNQA8OjklhfCpyJqPBgnb88Q+cvz1B3v0t7j7Vpz5FW1YxtKKxu7FV1Y69+u61KxbkJOndpqopzExXk5PgpALA6iuk0qxybLZUopsDZiosK0YWrsnThqix19AzrraNt2lXRqob2AXl9hsqqOlRW1eF/ulphis5dkqrcjGh29gOARVFMp9n4+tLkuDDFRoaYGwaYRRJjw3Tpuhxdui5H9W0D2lXeol2HW9Q74NbgiEevlDXqlbJGJceG6ZylqTpnSYqS49g0BQBWQjGdRoZhTBTT/MxYU7MAs1lWcqSytizUx87P1eG6Lu0qb1FpZbvco/6d/c/tqNVzO2q1MCNG5y5N1erCZDYiAoAFUEynUUfviLr7XZKkPDY+AQFnt9u0dH6Cls5P0A1uj0qPtWtXRYuOnOiWob+tR/3Ntkoty03UOUtTVZybIKeD9agAYAaK6TQany2VpALWlwLTKjT4bzv7u/pG9ObhVu2saFFj+6A8XkOlle0qrWxXZFiQVhcm69wlqVqQznpUAJhOFNNpVNXQI0mKiQxWUmyYuWGAOSw+OlSXrsvRJWuzVd82oJ3lLdp9uFV9g24NDI9q+75Gbd/XqJS4MJ27NFXri9IUH835qAAQaBTTaXSs3r8jvyArllkYwAJsNpuyU6KUnRKlqzfn6vCJbu0qb9G+yna5PT61dg/rmddr9eyOWi2dn6CNy9K0bGEit/oBIEAoptOkd9Ct1q4hSVIeG58Ay3HY7SpakKCiBQkadvnXo+4sb9bRkz0yDOnQ8U4dOt6pqPAgrV+apvOWpSktIcLs2AAwq1BMp0kV60uBGSMsxKkNxWnaUJymtp5h7TjYpDcOtai736X+oVH9ac9J/WnPSS3MiNF5y9K0elGyQoP54xQAzhZ/kk6T8Y1PEaFOpScxywLMFMmxYbpqY66u3LBA5bWdeu1Asw5Ud8jrM962q79KawuTdV5xOhumAOAsUEynyXgxzcuMlZ2/tIAZx263qTg3UcW5ieoddGtXeYteO9Cklq4hudxevXagWa8daFZGYoTOK07TOUtTFRUebHZsAJhRKKbTYGjEo/q2AUmcXwrMBjERwbpkbbYuXpOl6sZevX6gWXuOtso96lNjx6Aee7laT75So5L8JG0sTtPiefFmRwaAGYFiOg2qG3tljP04n/WlwKxhs9mUlxmrvMxYfeIDedpzpFWvH2zW8aY+eX2G9h5t096jbYqPDtHGZem6YuNCBXHDBADeFcV0Gozfxg8OsisnJcrcMAACIizEqU3LM7RpeYYa2ge042Czdpa3aGB4VF19Lj37uv8xqMW5CdpckqGl8xNkt9NSAeDtKKbToHLsYP3c9BjOPwTmgMykSF17QZ4+uilX+6s79PqBJlXUdskwpAPVnTpQ3anEmFBtLsnQhuI01qICwBiKaYC5R72qbeqTxG18YK4Jctq1elGyVi9KVs+gW28eadOfdp1Q76BbHb0jevKVGj3zeq1WL0rWlhUZ7OgHMOdRTAOsttm/1kyS8jPZ+ATMVYkxobr+0kJdvDpTew636uV9jaqs75HH69OuihbtqmhRdkqktqzI1NrFKQoJcpgdGQCmHcU0wI6NrS912G1akEExBeY6p8OuNYUpWlOYoob2AW0va9TO8ha53F6dbB3Qr/54VE+8XK31RWnavCJDqfHhZkcGgGlDMQ2w8Sc+zUuNYgYEwCkykyJ1w0UF+timXO2qaNH2fY1q7BjUkMujv+yt11/21mvxvDhtWZGpZQsT5LCzRh3A7EYxDSCvz6fqRtaXAnhvYSFObVmRqc0lGaqs79H2skaVHmuX12fo8IluHT7RrbioEJ2/PF0bl6UrJjLE7MgAEBAU0wA62Tog16hXkpRHMQXwPmw2mwqy41SQHafeAZdeO9CkV/Y3qbvfpe5+l555vVbPv3FCKwuStLkkQ/lZsWyWAjCrUEwDaPz8UpukPDY+AZiEmMgQXbF+vj54To72V3Vqe1mDDp/oltdnaM+RNu050qbMpEhdtDpLaxenKMjJbX4AMx/FNIDGi2lGUqQiQoPMDQNgRnLY7VpZkKSVBUlq7hzUK2VN2nGoWcMujxraB7T1D0f0u1drdMGKDJ1fksGZqABmNIppgPgMQ1UNvZKk/CxmSwGcvbSECH3iA3m6auMC7T7coj+/Va/mziH1Dbr1zOu1enFXndYvTdWFq7OUlhBhdlwAmDSKaYA0dwxqYHhUEhufAEytkGCHNi3P0HnL0lV+vEt/fuukDp/o1qjHp1f2+9elFucm6OLVWVqUE8c6VAAzBsU0QCrHZksliimAwLDbbCrOTVBxboLq2wb057dOandFq7w+QwdrOnWwplNZyX9bh8ojkQFYHX9KBcj4+tLkuDDFcrQLgADLSo7UTZct1n3/eq4uP3eeIsP869rr2wb0i98f0b/9dKde3Hli4k4OAFgRM6YBYBjGRDHNz4w1NQuAuSU2MkRXbVygy87J0a5y/zrUlq4h9Q649fRrx/XizhNaX5SmC1dn8VQpAJZDMQ2Ajt4Rdfe7JHEbH4A5QoIcOr8kQxuXp+tQTaf+/Fa9jtR1y+3xaXtZo7aXNWr5wkRdtDpLBdmchwrAGiimATA+WyqxIx+Auew2m5YtTNSyhYk62dqvP79VrzcP+9eh7q/u0P7qDmWnROri1dlaXZjMOlQApuJPoACoauiRJMVGBispNszcMAAwJjslSp+9fLG+e8u5uuycHEWE+ucmTrYO6MEXD+vun+/W9rJGjXq8JicFMFcxYxoAx+rHzy/l9hgA64mLCtFHN+Xq8nPmaWd5s/78Vr1au4fV0TuiX790TM+/UatL1mTr/OUZCgl2mB0XwBxCMZ1ivYNutXYNSWJ9KQBrCwl2aPOKTG1anqHSyna9uPOE6tsG1Dvg1uMvV+v3u+p04apMXbAyU+E8vQ7ANKCYTrGqt68vZUc+gBnAbrdp9aJkrSpI0sGaTr2484Rqmvo0MDyqZ16v1Z/2nNSWFZm6cHWWonnkKYAAophOsfGNTxGhTqUn8UhAADOHbWyjVHFugo7WdevFXXU6UtetYZdXv99Vp7/srdemZRm6ZG224qI4nxnA1KOYTrHxYpqXGSs760sBzEA2m02F8+JVOC9eNY29enHnCR2o6ZR71Ke/7K3X9rIGbShK06XrctjgCWBKUUyn0NCIR/VtA5KkPI6JAjAL5GbE6EtXL9PJ1n69uKtOpUfb5PEaemV/k1470Ky1i1N02Tk5Sk/kDhGAs0cxnULVjb0yxn7MxicAs0l2SpT+9cqlau4c1B921WlXRat8hqFdFS3aXdGiFQVJuvycecpJjTI7KoAZjGI6hcZv4wcH2ZWTwh/OAGaftIQI3XT5Yn14w3z98c2Tev1gszxen0qPtav0WLuKcxN0+TnztDCTu0YAJo9iOoUqxw7Wz02P4ekpAGa1xNgw3XBxga5YP08v7Tmp7WWNco/6dLCmUwdrOlWYE6erNi1QbjoFFcDpoz1NEfeoV7VNfZK4jQ9g7oiNDNE1W/J03y3n6vJz5yksxD/fcaSuW9/+31L96KmDahhbew8A74cZ0ylS29wnr8+/wpRiCmCuiQoP1lUbF+iSNdnaVlqvl/ac1LDLq7KqDu2v6tDaJSn68Ib5SokLNzsqAAujmE6RY2PrSx12mxakR5sbBgBMEh7q1IfWz9eWFZn64+46/bW0QW6PT7srWvXWkTadV5ymK9bP5xxUAP8QxXSKjD/xaV5qlEKCeLY0gLktMixIV29eqAtXZ+nFnSf06v4meX3+Y6Z2HGrRlhUZ+uA5OTxJCsApWGM6Bbw+n6obWV8KAH8vNjJE119UoP+6eZ3WL02VzSZ5vD79+a163fnALj37+nENjXjMjgnAIiimU+Bk64Bco15JUh7FFADeISk2TDddvljfuGmtVhYkSZJcbq+ef+OE7nxgp/74Zt3En6MA5i6K6RQYP7/UJimPs/sA4F1lJEbo1o8U6T//aZWWLoiXJA2OePTk9hr9+8926eV9DfJ4fSanBGAWiukUGC+mGUmRiggNMjcMAMwA81Kj9ZWPL9ed15VM/IO+d8CtR/5cqbt/vltvHGqWz2e8z2cBMNtQTM+SzzBU1dArSSrgNj4ATEpBdpz+/ZMr9OWrlyk7JVKS1NE7ol/8/oj+4xdvqvRYmwyDggrMFezKP0vNHYMaGB6VJOVlcRsfACbLZrOpODdBSxfEq/RYu5557bhauobU3Dmk+58p17zUKF17QR6bS4E5gGJ6lirHZkslduQDwNmw22xavShZK/ITtau8Vc/tOK7OPpdOtPTrvx/dp1WLkvXx83OVGBtmdlQAAUIxPUvj60uT48IUG8mB0QBwthx2uzYUp2nt4hS9sr9Rz++o1eCIR3uPtml/VYcuWZulD67LUWgwf4UBsw1rTM+CYRgTxZTZUgCYWkFOuy5claV7//kcXbAyU3abTR6vTy/urNNd4xukWH8KzCoU07PQ0Tui7n6XJCk/M9bcMAAwS0WGBemTF+brGzetmThiqnfArV/8/oi+9fBeVTX0mBsQwJShmJ6F8dlSScpn4xMABFR6YoS+8vHl+vLVxUqND5cknWjp172P7NMDz5Wrs3fE5IQAzhYLdM7C+L/SYyODlcRifACYFsW5iVo8L17b9zXquR21GnJ5tOfI+PrTbF26NkchwQ6zYwI4A8yYnoVj9f4d+flZsbLZbCanAYC5w+mw68LVWbr3n9dp84oM2WyS2+PT82+c0N0P7tau8hbWnwIzEMX0DPUOutXaNSSJjU8AYJao8GDdcFGBvv6ZNVoyL06S1N3v0oMvHtZ//bpUNY297/MZAFgJxfQMVb19fSkbnwDAVJlJkfrKNcv1xY8WKyXOv7TqeFOfvv3rUv38hQp19bH+FJgJWGN6hsY3PkWEOpWeFGFuGACAbDabluclaumCeP21tEHPv3FCwy6Pdle0at+xdl26LkeXr59ndkwA74EZ0zM0XkzzMmNlZ30pAFiG02HXxWuyde8/r9P5JX9bf/rcjlrd+dOdeuNgkwzWnwKWRDE9A0MjHtW3DUhifSkAWFV0eLBuvLhA93x6jQpz/OtPu/pc+u+H39L3nzzA7X3AgiimZ6C6sVfj/9bO4/xSALC0rORIffXa5frCVUVKjAmVJJVVduj/PPSmtu2tl8/H7ClgFRTTMzB+Gz84yK6clChzwwAA3pfNZlNJfpL+65/X6cpNubLZJJfbq99sq9K3f106cRcMgLkopmegcuxg/dz0GDkd/BYCwEwRGuzUTR9aqns+s0bZKZGSpNrmPn3jV2/pd6/UyD3qNTkhMLfRqibJPepVbVOfJNaXAsBMNT8tWv/xqVX6+OaFCg6yy+sz9IfddfrPX+xRxYkus+MBcxbFdJJqm/vkHVuPRDEFgJnLYbfrkrXZ+uZNa7V0frwkqa1nWP/vsf36xYuH1T/kNjkhMPdQTCfp2Nj6UofdpgXp0eaGAQCctaTYMN328WW6+YrFigoPkiS9Ud6i//Pgm9pV3sLRUsA0ophO0vgTn+alRSkkyGFuGADAlLDZbFq3JFXf/tw6bShKkyQNDI/qwRcP63uP71dbz7DJCYG5gWI6CV6fT9WNY+tLeQwpAMw6kWFB+sxlhfq3a5creezRphUnuvWfD72pP+6uk8frMzkhMLtRTCfhZOuAXGM7NllfCgCzV+G8eH3jM2t0+bk5cthtcnt8evKVGn3z4b2qbe4zOx4wa1FMJ2H8/FKbpLxMDtYHgNksOMihqzbm6mufXq3csT0F9W0D+tb/7tVvt1VpxO0xOSEw+1BMJ2G8mGYmRyo8NMjcMACAaZGZFKm7rl+pT16Yr9BghwxD+sveev3HQ2/qQHWH2fGAWYViOgl1rf2SWF8KAHON3W7TBSsz9a3PrlVJXqIkqbPPpR/87qAe/tNRudwczA9MBYrpJCzLTVRsZLA2laSbHQUAYIL46FB94aPFuvUjRYqJDJYkvbq/Sff86i3WngJTgGI6CTdcXKDvfX6DMpMizY4CADDRyoIkffOmtVqZnyRJau0a0n/9ulS/33VCPh/nngJnimIKAMAZiAwL0r9+ZKk+fekihQQ55PUZeurV4/rub8vU0cu5p8CZoJgCAHCGbDabzluWrns+s3riaYCV9T362ta3tPtwi8npgJmHYgoAwFlKiQvXv39yha44d55sNmnY5dHPnz+sn79QoaERjpUCThfFFACAKeB02PWRjQv0759cocSYUEnS7opWfW3rnonjBgG8N4opAABTKC8zVl//zBqduzRVktTZN6Lv/Gafnnq1hkeaAu+DYgoAwBQLC3Hqs5cv1r98eInCQ5wyDOn3u+p07yOlaukaMjseYFkUUwAAAmRNYYq+cdMaLcqOlSTVNvfrnl/u0av7G2UYHCsF/D2KKQAAARQfHaqvXluiq8/PlcNuk3vUp4f/dEw/fvqQ+ofcZscDLIViCgBAgNntNl26Lkf/342rlBofLkkqq+rQf/5ij8qPd5qcDrAOiikAANMkJzVKX/v0am0uyZAk9Q669b0nDug3f6nUqMdrcjrAfBRTAACmUUiQQzdcXKAvfqxYUeFBkqRtpQ36xq/2qrFj0OR0gLkopgAAmGD5wkR946a1Ks5NkCQ1dgzqW/+7V6XH2k1OBpiHYgoAgEliIoL1pY8V65MX5stht8nl9ur+Zw7pmdeOy8eufcxBFFMAAExks9l0wcpM3XFdiWIigiVJL+w8oR/97iCPM8WcQzEFAMAC8jJj9Z//tFoL0qMlSQdqOvWt/92r5k7WnWLuoJgCAGARcVEhuvO6FdpQnCZJauka0jcf3quyKtadYm6gmAIAYCFBTrs+fekiXX+Rf93piNurHz11SM/tqGXdKWY9iikAABZjs9m0ZUWm/u0TJYoeO1LquR21uv/pQxp2se4UsxfFFAAAi8rP8q87nZcaJcn/tKhv/e9etXQNmZwMCAyKKQAAFhYfHaq7rl+h9UtTJUnNnUP65sNv6UB1h8nJgKlHMQUAwOKCnA595rJCXfeBPNltNg27vPrh7w7qhTdYd4rZhWIKAMAMYLPZ9IFVWfrqtcsVGRYkQ9Izr9fqp8+Us+4UswbFFACAGWRRTpy+9k+rlZPiX3daWtmub/+6VK3drDvFzEcxBQBghkmI8a87PWdJiiSpqWNQ3/zVXh063mlyMuDsUEwBAJiBgoMc+uzli3XtBf51p0Muj77/xAH9ftcJGaw7xQxFMQUAYIay2Wy6aHWWbr9m2cS606dePa6fPlehETfrTjHzUEwBAJjhCufF6z8/tUrZyZGSpL1H23TvI/vUM+AyORkwORRTAABmgcTYMN11w0qtXexfd1rfNqB7HylVW8+wycmA00cxBQBglggJcujmKxbrinPnSZLae0Z07yOlamgbMDcYcJoopgAAzCI2m00f2bhA116QJ0nqHXDrvx/dp+rGXpOTAe+PYgoAwCx00eos3XRZ4cSO/f/7WJnKazlOCtZGMQUAYJZaX5SmW69aKqfDLveoTz948qDeOtpmdizgXVFMAQCYxUrykvSVjy9TaLBDXp+hB54t1yv7G82OBfxDFFMAAGa5RTlxuuO6komzTv/3T8c4iB+WRDEFAGAOmJcarbuuX6H46BBJ/oP4n9xeQzmFpVBMAQCYI9ISInT39SuVlhAuSfrTnpP65R+PyuvzmZwM8KOYAgAwh8RHh+rOT65QTmqUJGnHwWb99NkKjXq8JicDKKYAAMw50eHBuuMTJVqUHStJ2lfZru8/eVDDLo+5wTDnUUwBAJiDwkKcuu3jy1SSlyhJOlLXrf/7WJn6h9wmJ8NcRjEFAGCOCnI69K8fWar1RamSpNrmfv33o/vU1TdicjLMVRRTAADmMIfdrk9/sFAXrc6SJDV3DuneR0rV0jVkcjLMRRRTAADmOLvNpmu2LNRVGxdIkjr7XLr3kVLVtfSbnAxzDcUUAADIZrPp8nPn6YaL8mWT1D80qu/+dp+Onew2OxrmEIopAACYsHlFpm7+0BI57DYNu7z63hMHtL+6w+xYmCMopgAA4BRrF6foix8rVrDTrlGPTz9+6pBKj7WZHQtzAMUUAAC8Q9GCBN1+7XKFhzjlMwz97PkKHanjtj4Ci2IKAAD+obzMWN1+7XKFBDnk8Rr60VMH2RCFgKKYAgCAdzU/LVq3XrVUDrtNI26v/ueJ/Wrt5igpBAbFFAAAvKel8xN00+WFsknqGxrV9x7fr94Bl9mxMAtRTAEAwPtatzhVn/hAniSpvWdE33vigIZGPCanwmxDMQUAAKflA6uydPm58yRJ9W0D+uFTBzXq8ZobCrMKxRQAAJy2j5w3X5uWp0uSKut79MBzFfL6fCanwmxBMQUAAKfNZrPphosKtDI/SZJUVtWhX790TIZhmJwMswHFFAAATIrdbtPNH1qsRdmxkqTXDjTr6deOmxsKswLFFAAATFqQ06EvfLRY2SmRkqTf76rTX96qNzkVZjqKKQAAOCNhIU7d9vHlSo4NkyT99q9V2lXRYnIqzGQUUwAAcMZiIoL1lWuXKyYiWJK09fdHdOh4p8mpMFNRTAEAwFlJjg3TbR9fprAQh7w+Q/c/c0g1Tb1mx8IMRDEFAABnLTslSl/8aLGcDrvcoz59/4kDauoYNDsWZhiKKQAAmBIF2XH6lw8vkc0mDY549L0n9qurb8TsWJhBKKYAAGDKrMhP0qcuWSRJ6upz6f89vl8Dw6Mmp8JMQTEFAABTauOydH100wJJUnPnkL7/5AG53Dy6FO+PYgoAAKbcB9fl6AOrMiVJx5v6dP+zh+Tx8uhSvDeKKQAAmHI2m03XXpCndYtTJEnlx7u09Q9H5OPRpXgPFFMAABAQdptNn7msUEsXxEuSdle06rd/qZRBOcW7oJgCAICAcTrsuvXKIi1Ij5YkvbSnXk9trzY5FayKYgoAAAIqJNihL1+9TGkJ4ZKk//3DYVXUdpmcClZEMQUAAAEXGRak2z6+TBFhQTIM6YFny9U76DY7FiyGYgoAAKZFYkyYbr5isSSpd9CtB1+okM/HelP8DcUUAABMm5L8JF25KVeSdPhEt36/64S5gWApFFMAADCtbvzg4onNUM/uqNWxk90mJ4JVUEwBAMC0CnLadetVRQoPccowpJ89X6G+IdabgmIKAABMkBQbps9cVihJ6hlw66EXDnP4PiimAADAHCvyk/SBlf7HlpbXdumPu+tMTgSzUUwBAIBprt68UDmpUZKkZ16rVVVDj7mBYCqKKQAAME2Q065bPrxEYSEO+QxDDzxXoYHhUbNjwSQUUwAAYKrkuHD906X+9abd/S794sXDMlhvOidRTAEAgOlWL0rW5pIMSdKBmk69tKfe5EQwA8UUAABYwrUXLFR2cqQk6alXa1TT2GtyIkw3iikAALCEIKdDt1y5VCHBDnl9/vWmgyOsN51LKKYAAMAyUuLD9alLCiRJnX0j2vr7I6w3nUMopgAAwFLWLU7VxmXpkqSyqg5tK20wORGmC8UUAABYznUfyFNmUoQk6YmXq1Xb3GdyIkwHiikAALCc4KCx9aZB4+tNyzU04jE7FgKMYgoAACwpLSFCN1ycL0lq7xnRr/7IetPZjmIKAAAs69yladpQlCZJ2nusXdvLGk1OhECimAIAAEv75IX5Sk/0rzd97K9VqmvpNzkRAoViCgAALC0k2KFbPrxEwU67PF5DP32uXMMu1pvORhRTAABgeRlJkfrkhf71pm3dw3r4T0dZbzoLUUwBAMCMsKE4TecsSZEk7TnSptcONJmcCFONYgoAAGYEm82mGy4uUGp8uCTpN9uqVN82YHIqTCWKKQAAmDFCg5265cqlCnLaNerx6afPlmvEzXrT2YJiCgAAZpSs5Eh94gN5kqSWriG9uLPO5ESYKhRTAAAw42xalq7i3ARJ0p/fOqm27iGTE2EqUEwBAMCMY7PZdO0FeXLYbfJ4DT3+crXZkTAFKKYAAGBGSo0P14WrsiRJZVUdqjjRZXIinC2KKQAAmLEuP3eeosODJEmPbauS1+czORHOBsUUAADMWOGhTl21KVeS1NgxqFfKONt0JqOYAgCAGW1DUZpyUqIkSc++flwDw6MmJ8KZopgCAIAZzW63TRwfNTji0XOv15qcCGeKYgoAAGa8/KxYrSlMliRtL2tUQztPhJqJKKYAAGBWuPr8hQp22uUzDP12W5UMwzA7EiaJYgoAAGaFhJhQXbouR5J0pK5b+6s6TE6EyaKYAgCAWeOStdmKjw6RJD32cpVGPRwfNZNQTAEAwKwREuTQxzcvlCS194zoL3vrTU6EyaCYAgCAWWX1omTlZcZIkl7YeUI9Ay6TE+F0UUwBAMCsYrPZdN0H8mWT5HJ79dQrNWZHwmmimAIAgFknJzVK5y1LkyS9Ud6i4019JifC6aCYAgCAWekjG3MVFuKQJP1mW6V8HB9leRRTAAAwK8VEBOuKc+dLko439enNilaTE+H9UEwBAMCs9YFVmUqJD5ckPflKtUbcHpMT4b1QTAEAwKzldNj1iQv8x0f1DLj1h911JifCe6GYAgCAWa04N1FFCxIkSX96s17tPcMmJ8K7oZgCAIBZ79oLFspht8nj9emJ7dVmx8G7oJgCAIBZLy0hQheszJQklR5r15G6bpMT4R+hmAIAgDnhQ+vnKTIsSJL0221V8vp8JifC36OYAgCAOSE8NEhXbVogSWpoH9BrB5pNToS/RzEFAABzxsbidGUlR0qSnnntuAZHRk1OhLejmAIAgDnDbrfpug/kSZIGhkf13I5akxPh7SimAABgTinIjtOqRcmSpJdLG9XYMWhyIoyjmAIAgDnn4+fnKshpl88w9Nhfq2QYhtmRIIopAACYgxJjw3TJmmxJUkVtlw5Ud5qcCBLFFAAAzFEfXJejuKgQSdJjL1dp1MPxUWajmAIAgDkpJNihq8/PlSS1dQ9rW2m9yYlAMQUAAHPW2sUpWpgRI0l64Y0T6h1wmZxobqOYAgCAOctms+kTY8dHjbi9euZ1jo8yE8UUAADMafPTorV+aaokaWd5iwaGOXTfLBRTAAAw5108tkPf4/Vpx0EeVWoWiikAAJjzMpMjtTDTv9b0lbJG+TjX1BQUUwAAAEmbSzIkSW09wzp8osvkNHOTZYppbW2tVqxYYXYMAAAwR60qSFZkWJAkafu+RpPTzE2WKKbDw8P6zne+o5CQELOjAACAOSrIadd5xWmSpP3VHerqGzE50dxjiWL67W9/W7feeqvCwsLMjgIAAOawTSUZskkyDOm1A01mx5lznNP1hR5//HH9+te/PuW1X/ziF3rllVe0aNEiFRUVTVcUAACAfyg5NkxLFsSr/HiXXj3QpMvPnSenwxLzeHPCtBXTa665Rtdcc807Xn/++edlt9v10ksvqb29XZ/97Gf10EMPTVcsAACAU2wuyVD58S71Dri1v6pDqxYlmx1pzpi2YvpuHn300Ykfb9myhVIKAABMtSw3UfHRIerqc2l7WSPFdBqd8dy02+3W5ZdfrjfffHPiNZfLpbvvvlurVq3Shg0btHXr1ikJCQAAMF3sdps2LUuXJB2p61ZL15DJieaOM5oxdblcuv3221VVVXXK69/97ndVXl6uhx9+WE1NTbrzzjuVnp6uSy655LQ+78svvzzpLHa7TXa77V0/7hhbF+JgfYhpGAPzMQbmYwysgXEw30wZg80rM/X8Gyfk9Rl69UCTPnlhvtmRpoyVx2DSxbS6ulq33367jL97IsLQ0JCefPJJPfjgg1qyZImWLFmiqqoqPfroo6ddTM9EfHyEbLZ3L6bjoqPZ8W82xsB8jIH5GANrYBzMZ/UxiIuL0LqiNL1xoElvHGzW5z5SrJAgh9mxppQVx2DSxXTPnj1au3atbrvtNi1fvnzi9aNHj8rj8aikpGTitZUrV+qBBx6Qz+eT3R6YVt7VNfi+M6bR0WHq6xuW1+sLSAa8N8bAfIyB+RgDa2AczDeTxuC8olS9caBJA8OjeumN4zpv7Pb+TGfGGMTFRZzW+yZdTK+77rp/+Hp7e7vi4uIUHBw88VpiYqJcLpd6enoUHx8/2S91Wnw+Qz7f+z/P1uv1yeOx9gUw2zEG5mMMzMcYWAPjYL6ZMAZ5GTFKSwhXc+eQtu1t0DlLUs2ONKWsOAZTNo05PDx8SimVNPFzt9s9VV8GAABgWthsNp2/PEOSVNvcp7qWfpMTzX5TVkxDQkLeUUDHfx4aGjpVXwYAAGDarC9KVbDTX5e2lzWYnGb2m7JimpKSou7ubnk8nonX2tvbFRoaqujo6Kn6MgAAANMmPDRIaxanSJJ2H27V0IjnfX4FzsaUFdPCwkI5nU7t379/4rXS0lIVFRUFbOMTAABAoG0u8d/Od4/6tLO82eQ0s9uUNcawsDBdeeWVuueee3Tw4EFt27ZNW7du1Y033jhVXwIAAGDazU+L1rzUKEnS9rLGdxyZiakzpVOZd911l5YsWaJPfepT+vrXv64vfOELuuiii6bySwAAAEy78VnT5s4hHTvZY26YWeyMnvw07tixY6f8PCwsTN/5znf0ne9856xCAQAAWMmaxSl6/OVqDbk82l7WqEU5cWZHmpVY/AkAAPA+QoIcOrfIf47pvsp29Q64TE40O1FMAQAATsP47Xyvz9BrB9kEFQgUUwAAgNOQlhChRdmxkqTX9jee1pMnMTkUUwAAgNO0eUWmJKmzz6WDNZ0mp5l9KKYAAACnqSQvUTER/keuby9rNDnN7EMxBQAAOE1Oh10bl6VLksqPd6q9Z9jkRLMLxRQAAGASNi1Pl80mGZJe2c+s6VSimAIAAExCfHSoli9MlCS9fqBZox6fyYlmD4opAADAJI0fHTUwPKrSY20mp5k9KKYAAACTtHh+vJJjwySxCWoqUUwBAAAmyW6zaVOJfxNUVUOvGtoGTE40O1BMAQAAzsCGojQ5Hf4qtZ1NUFOCYgoAAHAGosKDtXpRkiRpV3mLRtwekxPNfBRTAACAM7S5xP8kqBG3V7srWk1OM/NRTAEAAM5Qbka0MpMiJfk3QRmGYXKimY1iCgAAcIZsNps2r/AfHVXfNqCapj6TE81sFFMAAICzsG5xikKCHZKk7fsaTE4zs1FMAQAAzkJYiFPnLkmVJL11tE39Q26TE81cFFMAAICzNP4kKI/X0I5DzSanmbkopgAAAGcpMzlSCzNjJEmvljXJxyaoM0IxBQAAmALjs6ZtPcM6XNtlcpqZiWIKAAAwBVYVJCsyLEiS/+goTB7FFAAAYAoEOe06rzhNkrS/ukNdfSMmJ5p5KKYAAABTZFNJhmySDEN6dX+T2XFmHIopAADAFEmODdOSBfGSpNcONsnj9ZmcaGahmAIAAEyh8U1QvQNuHavvMTfMDEMxBQAAmEJL58fL6bBJkipP9pgbZoahmAIAAEyhIKdDC9KiJUlVDT3mhplhKKYAAABTLC8rVpJU09THOtNJoJgCAABMsYKxYjrq8elEc7+5YWYQiikAAMAUy82Ikc2/zFSV3M4/bRRTAACAKRYW4lR2SpQkqZKd+aeNYgoAABAA+ZmxkqSqhl75fIa5YWYIiikAAEAA5I+tMx12edTQPmBumBmCYgoAABAAeVkxEz/mdv7poZgCAAAEQHR4sNISwiVJlQ29JqeZGSimAAAAATJ+O7+yvkeGwTrT90MxBQAACJDxDVB9g261dQ+bG2YGoJgCAAAEyPiMqSQdY53p+6KYAgAABEhCTKgSokMkSVUU0/dFMQUAAAig8VlTZkzfH8UUAAAggPLGimlH74i6+kbMDWNxFFMAAIAAKnjbOtPKhh7TcswEFFMAAIAASo0PV1R4kCSpqp7zTN8LxRQAACCAbDbbxLFRzJi+N4opAABAgI2vM21sH9TA8Ki5YSyMYgoAABBgb19nWsWs6buimAIAAARYVnKkQoMdkvyPJ8U/RjEFAAAIMLvdpoUZMZKkSjZAvSuKKQAAwDQYP2i/rqVfI26PuWEsimIKAAAwDcaLqc8wVNPUZ24Yi6KYAgAATIP5aVFyOvzVq/Jkj7lhLIpiCgAAMA2CnA4tSIuSxM78d0MxBQAAmCb52bGSpJqmPnm8PnPDWBDFFAAAYJqMPwFq1OPTieZ+c8NYEMUUAABgmuRmxMhm8/+Yx5O+E8UUAABgmoSFOJWd4l9nykH770QxBQAAmEbjjyetauiVz2eYG8ZiKKYAAADTKG9snemwy6OG9gFzw1gMxRQAAGAa5WXFTPyY2/mnopgCAABMo+jwYKUlhEuimP49iikAAMA0G388aWVDrwyDdabjKKYAAADTbLyY9g261dY9bG4YC6GYAgAATLPxg/Yl6Ri38ydQTAEAAKZZQkyoEqJDJUlVFNMJFFMAAAAT5I/tzmfG9G8opgAAACYYX2fa0Tuirr4Rc8NYBMUUAADABOPFVJIqG3pMy2ElFFMAAAATpMaHKyo8SJJUVd9rchproJgCAACYwGazTezO56B9P4opAACAScZv5zd2DGpgeNTcMBZAMQUAADDJ29eZVrHOlGIKAABglqzkSIUGOyRxO1+imAIAAJjGbrdpYab/PNNKNkBRTAEAAMw0vgGqrqVfI26PuWFMRjEFAAAw0fg6U59hqKapz9wwJqOYAgAAmGh+WrScDn8lqzzZY24Yk1FMAQAATBTktGtBerQkduZTTAEAAEyWn+XfAFXT1KdRj8/kNOahmAIAAJhsfJ3pqMenupZ+c8OYiGIKAABgstz0GNls/h9XzuHb+RRTAAAAk4WFOJWTEiVpbh+0TzEFAACwgPHb+VUNvfL5DHPDmIRiCgAAYAF5YwftD7s8amgfMDeMSSimAAAAFpA3tjNfmru38ymmAAAAFhAdHqy0hHBJFFMAAACYrGBsnWllQ68MY+6tM6WYAgAAWETeWDHtG3SrtXvY3DAmoJgCAABYxPiMqTQ3b+dTTAEAACwiPjpUCdGhkqQqiikAAADMNH6e6TGKKQAAAMyUP3ZsVEfviLr6RkxOM70opgAAABaS//Z1pg09puUwA8UUAADAQlLjwxUdHiRJqqrvNTnN9KKYAgAAWIjNZps4Nmqu7cynmAIAAFhMfmasJKmxY1ADw6PmhplGFFMAAACLefs607l0bBTFFAAAwGKykiMVGuyQNLc2QFFMAQAALMZut2lhpv/YqMo5tAGKYgoAAGBB448nrWvp14jbY26YaUIxBQAAsKC8sQ1QPsNQTVOfuWGmCcUUAADAguanRcvp8Fe1ypM95oaZJhRTAAAACwpy2rUgPVqSVDVHNkBRTAEAACxq/NiomqY+jXp85oaZBhRTAAAAi8rP8u/MH/X4VNfSb3KawKOYAgAAWFRueozsNpsk6Vh9t8lpAo9iCgAAYFFhIU5lJkVIkurbBkxOE3gUUwAAAAsLGXsClM9nmJwk8CimAAAAsASKKQAAACyBYgoAAABLoJgCAADAEiimAAAAsASKKQAAACyBYgoAAABLoJgCAADAEiimAAAAsASKKQAAACyBYgoAAABLoJgCAADAEiimAAAAsASKKQAAACyBYgoAAABLoJgCAADAEiimAAAAsASKKQAAACzBZhiGYXYIAAAAgBlTAAAAWALFFAAAAJZAMQUAAIAlUEwBAABgCRRTAAAAWALFFAAAAJZAMQUAAIAlUEwBAABgCRRTAAAAWMKsKKYul0t33323Vq1apQ0bNmjr1q3v+t7Dhw/r6quv1rJly/TRj35U5eXl05h09prMGNxyyy0qKCg45b/t27dPY9rZze126/LLL9ebb775ru/hOgis0xkDroPAaG1t1Re/+EWtWbNG5513nu699165XK5/+F6ug8CZzDhwLQRGXV2dbrrpJpWUlOj888/XQw899K7vtdK14DTtK0+h7373uyovL9fDDz+spqYm3XnnnUpPT9cll1xyyvuGhoZ0880364orrtB///d/67e//a3++Z//WX/5y18UHh5uUvrZ4XTHQJJqamp033336Zxzzpl4LSYmZjrjzloul0u33367qqqq3vU9XAeBdTpjIHEdBIJhGPriF7+o6OhoPfroo+rt7dXdd98tu92uO++885T3ch0EzmTGQeJaCASfz6ebb75ZRUVFeuaZZ1RXV6evfOUrSklJ0RVXXHHKey13LRgz3ODgoFFUVGTs3r174rX777/fuP7669/x3ieffNLYsmWL4fP5DMMwDJ/PZ1x44YXGU089NW15Z6PJjIHL5TIKCwuN48ePT2fEOaGqqsr40Ic+ZFxxxRVGfn7+KePxdlwHgXO6Y8B1EBjV1dVGfn6+0d7ePvHaCy+8YGzYsOEd7+U6CJzJjAPXQmC0trYaX/rSl4z+/v6J12699Vbja1/72jvea7VrYcbfyj969Kg8Ho9KSkomXlu5cqUOHDggn893ynsPHDiglStXymazSZJsNptWrFih/fv3T2fkWWcyY3D8+HHZbDZlZWVNd8xZb8+ePVq7dq0ef/zx93wf10HgnO4YcB0ERlJSkh566CElJiae8vrAwMA73st1EDiTGQeuhcBITk7W97//fUVGRsowDJWWluqtt97SmjVr3vFeq10LM/5Wfnt7u+Li4hQcHDzxWmJiolwul3p6ehQfH3/KexcuXHjKr09ISHjfW254b5MZg+PHjysyMlJ33HGH9uzZo9TUVH3hC1/Qpk2bzIg+q1x33XWn9T6ug8A53THgOgiM6OhonXfeeRM/9/l8euSRR7Ru3bp3vJfrIHAmMw5cC4G3ZcsWNTU1afPmzbr44ovf8XGrXQszfsZ0eHj4lEIkaeLnbrf7tN779+/D5ExmDI4fP66RkRFt2LBBDz30kDZt2qRbbrlFhw4dmra8cx3Xgfm4DqbHfffdp8OHD+u22257x8e4DqbPe40D10Lg/fCHP9QDDzygI0eO6N57733Hx612Lcz4GdOQkJB3/OaN/zw0NPS03vv378PkTGYM/vVf/1U33HDDxML2RYsWqaKiQk888YSKioqmJ/Acx3VgPq6DwLvvvvv08MMP63/+53+Un5//jo9zHUyP9xsHroXAG/99dLlc+upXv6o77rjjlCJqtWthxs+YpqSkqLu7Wx6PZ+K19vZ2hYaGKjo6+h3v7ejoOOW1jo4OJScnT0vW2WoyY2C329+x23LBggVqbW2dlqzgOrACroPA+uY3v6lf/vKXuu+++/7hrUuJ62A6nM44cC0ERkdHh7Zt23bKawsXLtTo6Og71vpa7VqY8cW0sLBQTqfzlEW6paWlKioqkt1+6re3bNkylZWVyTAMSf4jLfbt26dly5ZNZ+RZZzJj8O///u+66667Tnnt6NGjWrBgwXREhbgOrIDrIHB+/OMf67HHHtP3vvc9XXbZZe/6Pq6DwDrdceBaCIyGhgZ9/vOfP6Xgl5eXKz4+/pR9H5L1roUZX0zDwsJ05ZVX6p577tHBgwe1bds2bd26VTfeeKMk/8zdyMiIJOmSSy5RX1+fvv3tb6u6ulrf/va3NTw8rEsvvdTMb2HGm8wYbNmyRS+88IKeffZZ1dXV6cc//rFKS0t1/fXXm/ktzHpcB+bjOgi8mpoa/eQnP9HnPvc5rVy5Uu3t7RP/SVwH02Uy48C1EBhFRUVasmSJ7r77blVXV+vVV1/Vfffdp3/5l3+RZPFrwZRDqqbY0NCQcccddxjLly83NmzYYPzyl7+c+Fh+fv4pZ3EdOHDAuPLKK42ioiLjYx/7mFFRUWFC4tlnMmPwxBNPGBdddJGxdOlS4yMf+YixZ88eExLPbn9/hibXwfR7vzHgOph6P/vZz4z8/Px/+J9hcB1Ml8mOA9dCYLS0tBi33nqrsWLFCmP9+vXGT3/604mzSq18LdgMY2zuFgAAADDRjL+VDwAAgNmBYgoAAABLoJgCAADAEiimAAAAsASKKQAAACyBYgoAAABLoJgCAADAEiimAAAAsASKKQDMAlu2bNGPfvQjs2MAwFmhmAIAAMASKKYAAACwBIopAEyB/v5+/cd//IfWrVunlStX6sYbb9ShQ4ckST/60Y/0iU98Qvfff7/Wrl2rVatW6a677tLAwMDEr+/p6dHXv/51bdq0ScXFxbr22mv15ptvnvI1Xn/9dV1zzTVatmyZNm7cqP/5n/+R1+ud+Hh7e7s+//nPa/ny5Vq7dq3uvffeUz4OAFZHMQWAs2QYhj73uc+pvr5eP/vZz/TEE09o+fLl+sQnPqHDhw9Lkg4dOqQdO3Zo69atuv/++/XWW2/py1/+siTJ6/XqM5/5jPbu3av77rtPTz/9tPLz83XTTTfp4MGDkqSysjLdfPPNWrlypZ5++ml961vf0mOPPaaf/OQnEzl+97vfafXq1XrhhRf0b//2b/rVr36lZ555Ztp/PwDgTNkMwzDMDgEAM9muXbv06U9/Wrt371ZsbOzE69dff70yMzOVkZGhn//859q2bZtSUlIkSa+99po+97nP6Y9//KPq6+t1880364UXXlB+fr4kf9m96qqrlJ2drR/84Af6yle+oqamJj322GMTn/+ll15SZ2enrrvuOm3ZskXFxcX6/ve/P/HxD33oQ1qxYoXuueee6fhtAICz5jQ7AADMdBUVFTIMQ5s3bz7ldbfbLZfLpYyMDM2bN2+ilErSihUrJEmVlZWqr69XVFTURCmVJJvNplWrVmnHjh0T71u/fv0pn//iiy8+5efz5s075ecxMTFyuVxn/f0BwHShmALAWfL5fIqMjNTTTz/9jo8FBwfrd7/7nYKCgk55fXztp8Ph0LvduDIMQ06n/4/p8f99Lw6H4x9+DgCYKVhjCgBnKT8/XwMDAxodHVVOTs7Efw8++KD++te/SpJqa2vV398/8WvKysokSYsXL1ZBQYH6+/tVWVk58XHDMFRaWqqFCxdKknJzcyc2U417+OGHdfXVVwf62wOAaUMxBYCzdN5556mwsFC33Xabdu/erbq6Ot177716+umnlZubK0kaGhrSHXfcocrKSu3cuVPf+MY39MEPflAZGRnasGGDCgsLdfvtt2vPnj2qqanRN77xDVVWVupTn/qUJOmzn/2s9u/frx/84Ac6ceKEXn31Vf3kJz/R+eefb+J3DgBTi1v5AHCWHA6Htm7dqvvuu09f/vKXNTw8rNzcXP34xz/WOeeco7179yotLU2FhYX65Cc/KYfDoSuuuEJf/epXT/n13/nOd/T5z39ebrdbS5cu1a9+9SstX75cklRYWKj7779fP/zhD/Xggw8qOTlZN954o2655RYTv3MAmFrsygeAAPvRj36kZ555Ri+//LLZUQDA0riVDwAAAEugmAIAAMASuJUPAAAAS2DGFAAAAJZAMQUAAIAlUEwBAABgCRRTAAAAWALFFAAAAJZAMQUAAIAlUEwBAABgCRRTAAAAWML/D+LDlzEOREn6AAAAAElFTkSuQmCC",
-      "text/plain": [
-       "<Figure size 800x550 with 1 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    }
-   ],
+   "outputs": [],
    "source": [
-    "df_hist['learning_rate'].plot(logy=True)"
+    "# df_hist['learning_rate'].plot(logy=True)"
    ]
   },
   {
@@ -1019,14 +1309,221 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 44,
    "metadata": {},
    "outputs": [],
-   "source": []
+   "source": [
+    "# df_res = pd.DataFrame(data)\n"
+   ]
   },
   {
    "cell_type": "code",
-   "execution_count": 28,
+   "execution_count": 84,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "sum\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "train/diff_sum      -0.474407\n",
+       "train/diff_mean     -0.383682\n",
+       "train/diff%_sum     -0.253703\n",
+       "train/diff%_mean    -0.253703\n",
+       "train/diff_std      -0.226926\n",
+       "train/diff%_std     -0.184180\n",
+       "train/diff_min      -0.082520\n",
+       "train/after_std      0.016230\n",
+       "train/before_max     0.026255\n",
+       "train/diff%_max      0.056223\n",
+       "train/diff_max       0.059453\n",
+       "train/after_mean     0.073286\n",
+       "train/after_max      0.082784\n",
+       "train/before_min     0.126469\n",
+       "train/after_min      0.176288\n",
+       "train/after_sum      0.182927\n",
+       "train/before_std     0.281097\n",
+       "train/diff%_min      0.342018\n",
+       "train/before_mean    0.388784\n",
+       "train/before_sum     0.455190\n",
+       "novelty              1.000000\n",
+       "Name: novelty, dtype: float64"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "df_res = pd.DataFrame(data)\n",
+    "for stat in ['mean', 'std', 'min', 'max', 'sum']:\n",
+    "    agg = getattr(np, stat)\n",
+    "    df_res[f'train/before_{stat}'] = df_res['train/before'].apply(lambda x: agg(x))\n",
+    "    df_res[f'train/after_{stat}'] = df_res['train/after'].apply(lambda x: agg(x)) \n",
+    "\n",
+    "    # df_res[f'before_{stat}'] = df_res['before'].apply(lambda x: agg(x))\n",
+    "    # df_res[f'after_{stat}'] = df_res['after'].apply(lambda x: agg(x))\n",
+    "\n",
+    "    # df_res[f\"diff_{stat}\"] = - df_res[f'before_{stat}'] + df_res[f'after_{stat}']\n",
+    "    # df_res[f\"diff%_{stat}\"] = df_res[f\"diff_{stat}\"] / df_res[f'before_{stat}'] * 100\n",
+    "\n",
+    "    df_res[f\"train/diff_{stat}\"] = -df_res[f'train/before_{stat}'] + df_res[f'train/after_{stat}']\n",
+    "    df_res[f\"train/diff%_{stat}\"] = df_res[f\"train/diff_{stat}\"] / df_res[f'train/before_{stat}'] * 100\n",
+    "\n",
+    "r = df_res.select_dtypes(include=np.number).corr()['novelty'].sort_values()\n",
+    "print(stat)\n",
+    "display(r)\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 89,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "only lesswrong\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/tmp/ipykernel_1883910/1966928096.py:2: FutureWarning: Downcasting object dtype arrays on .fillna, .ffill, .bfill is deprecated and will change in a future version. Call result.infer_objects(copy=False) instead. To opt-in to the future behavior, set `pd.set_option('future.no_silent_downcasting', True)`\n",
+      "  m = df_res.url.str.contains('lesswrong').fillna(False)\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "train/diff_sum      -0.247629\n",
+       "train/diff_mean     -0.159304\n",
+       "train/before_max    -0.077736\n",
+       "train/diff_std      -0.041644\n",
+       "train/diff%_sum     -0.034815\n",
+       "train/diff%_mean    -0.034815\n",
+       "train/before_min    -0.007299\n",
+       "train/after_min     -0.005681\n",
+       "train/diff%_std     -0.001165\n",
+       "train/diff_min       0.007247\n",
+       "train/after_std      0.111126\n",
+       "train/after_mean     0.168236\n",
+       "train/after_max      0.178731\n",
+       "train/before_std     0.202097\n",
+       "train/after_sum      0.261667\n",
+       "train/diff_max       0.264107\n",
+       "train/diff%_max      0.274232\n",
+       "train/diff%_min      0.318505\n",
+       "train/before_mean    0.390433\n",
+       "train/before_sum     0.418176\n",
+       "novelty              1.000000\n",
+       "Name: novelty, dtype: float64"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "without lesswrong\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "train/diff_sum      -0.524070\n",
+       "train/diff_mean     -0.289291\n",
+       "train/diff%_sum     -0.212006\n",
+       "train/diff%_mean    -0.212006\n",
+       "train/diff_std      -0.205211\n",
+       "train/diff%_max     -0.173305\n",
+       "train/diff_max      -0.153685\n",
+       "train/diff%_std     -0.135741\n",
+       "train/diff_min       0.065099\n",
+       "train/after_sum      0.075106\n",
+       "train/after_mean     0.077967\n",
+       "train/before_min     0.133288\n",
+       "train/after_std      0.148123\n",
+       "train/before_mean    0.209628\n",
+       "train/before_sum     0.217732\n",
+       "train/after_max      0.219580\n",
+       "train/before_max     0.319630\n",
+       "train/before_std     0.337727\n",
+       "train/after_min      0.384317\n",
+       "train/diff%_min      0.484895\n",
+       "novelty              1.000000\n",
+       "Name: novelty, dtype: float64"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "only new\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "train/diff_sum      -0.445037\n",
+       "train/diff_mean     -0.296686\n",
+       "train/diff_std      -0.175885\n",
+       "train/diff%_sum     -0.169818\n",
+       "train/diff%_mean    -0.169818\n",
+       "train/diff%_std     -0.149809\n",
+       "train/diff_min      -0.063278\n",
+       "train/after_std     -0.029953\n",
+       "train/before_max    -0.012773\n",
+       "train/after_mean     0.102790\n",
+       "train/after_max      0.116678\n",
+       "train/before_min     0.129897\n",
+       "train/diff%_max      0.131804\n",
+       "train/diff_max       0.133860\n",
+       "train/before_std     0.172009\n",
+       "train/after_min      0.238759\n",
+       "train/after_sum      0.306655\n",
+       "train/before_mean    0.357417\n",
+       "train/diff%_min      0.367850\n",
+       "train/before_sum     0.536616\n",
+       "novelty              1.000000\n",
+       "Name: novelty, dtype: float64"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "# also try with only new stuff, no  less wrong, only less wrong...\n",
+    "m = df_res.url.str.contains('lesswrong').fillna(False)\n",
+    "r = df_res[m].select_dtypes(include=np.number).corr()['novelty'].sort_values()\n",
+    "print('only lesswrong')\n",
+    "display(r)\n",
+    "\n",
+    "r = df_res[~m].select_dtypes(include=np.number).corr()['novelty'].sort_values()\n",
+    "print('without lesswrong')\n",
+    "display(r)\n",
+    "\n",
+    "r = df_res[~df_res.in_training].select_dtypes(include=np.number).corr()['novelty'].sort_values()\n",
+    "print('only new')\n",
+    "display(r)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 86,
    "metadata": {},
    "outputs": [
     {
@@ -1050,17 +1547,366 @@
        "  <thead>\n",
        "    <tr style=\"text-align: right;\">\n",
        "      <th></th>\n",
-       "      <th>before</th>\n",
-       "      <th>after</th>\n",
-       "      <th>len</th>\n",
-       "      <th>improvement%</th>\n",
-       "      <th>improvement</th>\n",
-       "      <th>novel</th>\n",
-       "      <th>learnable</th>\n",
-       "      <th>BS</th>\n",
+       "      <th>f</th>\n",
+       "      <th>train/diff_sum</th>\n",
+       "      <th>novelty</th>\n",
+       "      <th>in_training</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>22</th>\n",
+       "      <td>../samples/2025_lw_review-planecrash.md</td>\n",
+       "      <td>-339.481621</td>\n",
+       "      <td>0.933950</td>\n",
+       "      <td>False</td>\n",
        "    </tr>\n",
        "    <tr>\n",
+       "      <th>15</th>\n",
+       "      <td>../samples/2025_lw_2024-in-ai-predictions.md</td>\n",
+       "      <td>-297.344904</td>\n",
+       "      <td>0.789190</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>23</th>\n",
+       "      <td>../samples/2025_lw_the-field-of-ai-alignment-a...</td>\n",
+       "      <td>-288.831029</td>\n",
+       "      <td>0.925483</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>24</th>\n",
+       "      <td>../samples/2025_lw_the-intelligence-curse.md</td>\n",
+       "      <td>-265.925183</td>\n",
+       "      <td>0.688044</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>19</th>\n",
+       "      <td>../samples/2025_lw_my-agi-safety-research-2024...</td>\n",
+       "      <td>-261.054373</td>\n",
+       "      <td>0.756925</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>16</th>\n",
+       "      <td>../samples/2025_lw_comment-on-death-and-the-go...</td>\n",
+       "      <td>-239.929921</td>\n",
+       "      <td>0.750253</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>21</th>\n",
+       "      <td>../samples/2025_lw_preference-inversion.md</td>\n",
+       "      <td>-238.890490</td>\n",
+       "      <td>0.633321</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>17</th>\n",
+       "      <td>../samples/2025_lw_debating-buying-nvda-in-201...</td>\n",
+       "      <td>-229.822011</td>\n",
+       "      <td>0.526975</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>26</th>\n",
+       "      <td>../samples/2025_lw_the-subset-parity-learning-...</td>\n",
+       "      <td>-224.939272</td>\n",
+       "      <td>0.697946</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>8</th>\n",
+       "      <td>../samples/2024_lesswrong_slop.md</td>\n",
+       "      <td>-201.960423</td>\n",
+       "      <td>0.100000</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>18</th>\n",
+       "      <td>../samples/2025_lw_human-study-on-ai-spear-phi...</td>\n",
+       "      <td>-200.020488</td>\n",
+       "      <td>0.677458</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>12</th>\n",
+       "      <td>../samples/2024_openai_emails.md</td>\n",
+       "      <td>-194.183420</td>\n",
+       "      <td>0.700000</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>11</th>\n",
+       "      <td>../samples/2024_news_anthropic.md</td>\n",
+       "      <td>-176.985800</td>\n",
+       "      <td>0.500000</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>9</th>\n",
+       "      <td>../samples/2024_lw_by-default-capital-will-mat...</td>\n",
+       "      <td>-169.866388</td>\n",
+       "      <td>0.906388</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>../samples/2024_deliberative_alignment.md</td>\n",
+       "      <td>-159.913571</td>\n",
+       "      <td>0.600000</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>14</th>\n",
+       "      <td>../samples/2025_h5n1_report.md</td>\n",
+       "      <td>-156.453212</td>\n",
+       "      <td>0.750000</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>../samples/2024_anthropic_palintir.md</td>\n",
+       "      <td>-143.663636</td>\n",
+       "      <td>0.200000</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>6</th>\n",
+       "      <td>../samples/2024_gwern_reddit.md</td>\n",
+       "      <td>-142.434464</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>10</th>\n",
+       "      <td>../samples/2024_lw_the-plan-2024-update.md</td>\n",
+       "      <td>-136.823458</td>\n",
+       "      <td>0.785170</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>7</th>\n",
+       "      <td>../samples/2024_how_to_focus.md</td>\n",
+       "      <td>-135.387340</td>\n",
+       "      <td>0.500000</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>13</th>\n",
+       "      <td>../samples/2024_trump_appointment.md</td>\n",
+       "      <td>-128.444646</td>\n",
+       "      <td>0.300000</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>28</th>\n",
+       "      <td>../samples/lorem_ipsum.md</td>\n",
+       "      <td>-127.469837</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>True</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>../samples/2024_bob_fanfic2.md</td>\n",
+       "      <td>-125.559481</td>\n",
+       "      <td>0.400000</td>\n",
+       "      <td>True</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>5</th>\n",
+       "      <td>../samples/2024_gpt4_fake_paper.md</td>\n",
+       "      <td>-122.651137</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>29</th>\n",
+       "      <td>../samples/politics_is_the_mind_killer.md</td>\n",
+       "      <td>-116.203043</td>\n",
+       "      <td>0.500000</td>\n",
+       "      <td>True</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>25</th>\n",
+       "      <td>../samples/2025_lw_the-laws-of-large-numbers.md</td>\n",
+       "      <td>-107.898850</td>\n",
+       "      <td>0.540932</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>../samples/2024_bob_fanfic.md</td>\n",
+       "      <td>-106.220317</td>\n",
+       "      <td>0.300000</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>27</th>\n",
+       "      <td>../samples/2025_lw_what-s-the-short-timeline-p...</td>\n",
+       "      <td>-102.163854</td>\n",
+       "      <td>0.898161</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>20</th>\n",
+       "      <td>../samples/2025_lw_parkinson-s-law-and-the-ide...</td>\n",
+       "      <td>-97.869906</td>\n",
+       "      <td>0.677458</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>../samples/2024_arxiv_meh.md</td>\n",
+       "      <td>-73.838489</td>\n",
+       "      <td>0.150000</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                                                    f  train/diff_sum  \\\n",
+       "22            ../samples/2025_lw_review-planecrash.md     -339.481621   \n",
+       "15       ../samples/2025_lw_2024-in-ai-predictions.md     -297.344904   \n",
+       "23  ../samples/2025_lw_the-field-of-ai-alignment-a...     -288.831029   \n",
+       "24       ../samples/2025_lw_the-intelligence-curse.md     -265.925183   \n",
+       "19  ../samples/2025_lw_my-agi-safety-research-2024...     -261.054373   \n",
+       "16  ../samples/2025_lw_comment-on-death-and-the-go...     -239.929921   \n",
+       "21         ../samples/2025_lw_preference-inversion.md     -238.890490   \n",
+       "17  ../samples/2025_lw_debating-buying-nvda-in-201...     -229.822011   \n",
+       "26  ../samples/2025_lw_the-subset-parity-learning-...     -224.939272   \n",
+       "8                   ../samples/2024_lesswrong_slop.md     -201.960423   \n",
+       "18  ../samples/2025_lw_human-study-on-ai-spear-phi...     -200.020488   \n",
+       "12                   ../samples/2024_openai_emails.md     -194.183420   \n",
+       "11                  ../samples/2024_news_anthropic.md     -176.985800   \n",
+       "9   ../samples/2024_lw_by-default-capital-will-mat...     -169.866388   \n",
+       "4           ../samples/2024_deliberative_alignment.md     -159.913571   \n",
+       "14                     ../samples/2025_h5n1_report.md     -156.453212   \n",
+       "0               ../samples/2024_anthropic_palintir.md     -143.663636   \n",
+       "6                     ../samples/2024_gwern_reddit.md     -142.434464   \n",
+       "10         ../samples/2024_lw_the-plan-2024-update.md     -136.823458   \n",
+       "7                     ../samples/2024_how_to_focus.md     -135.387340   \n",
+       "13               ../samples/2024_trump_appointment.md     -128.444646   \n",
+       "28                          ../samples/lorem_ipsum.md     -127.469837   \n",
+       "3                      ../samples/2024_bob_fanfic2.md     -125.559481   \n",
+       "5                  ../samples/2024_gpt4_fake_paper.md     -122.651137   \n",
+       "29          ../samples/politics_is_the_mind_killer.md     -116.203043   \n",
+       "25    ../samples/2025_lw_the-laws-of-large-numbers.md     -107.898850   \n",
+       "2                       ../samples/2024_bob_fanfic.md     -106.220317   \n",
+       "27  ../samples/2025_lw_what-s-the-short-timeline-p...     -102.163854   \n",
+       "20  ../samples/2025_lw_parkinson-s-law-and-the-ide...      -97.869906   \n",
+       "1                        ../samples/2024_arxiv_meh.md      -73.838489   \n",
+       "\n",
+       "     novelty  in_training  \n",
+       "22  0.933950        False  \n",
+       "15  0.789190        False  \n",
+       "23  0.925483        False  \n",
+       "24  0.688044        False  \n",
+       "19  0.756925        False  \n",
+       "16  0.750253        False  \n",
+       "21  0.633321        False  \n",
+       "17  0.526975        False  \n",
+       "26  0.697946        False  \n",
+       "8   0.100000        False  \n",
+       "18  0.677458        False  \n",
+       "12  0.700000        False  \n",
+       "11  0.500000        False  \n",
+       "9   0.906388        False  \n",
+       "4   0.600000        False  \n",
+       "14  0.750000        False  \n",
+       "0   0.200000        False  \n",
+       "6   1.000000        False  \n",
+       "10  0.785170        False  \n",
+       "7   0.500000        False  \n",
+       "13  0.300000        False  \n",
+       "28  0.000000         True  \n",
+       "3   0.400000         True  \n",
+       "5   0.000000        False  \n",
+       "29  0.500000         True  \n",
+       "25  0.540932        False  \n",
+       "2   0.300000        False  \n",
+       "27  0.898161        False  \n",
+       "20  0.677458        False  \n",
+       "1   0.150000        False  "
+      ]
+     },
+     "execution_count": 86,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "main_metric = 'train/diff_sum'\n",
+    "df_res[['f', main_metric, 'novelty', 'in_training']].sort_values( main_metric)  "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 87,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
        "      <th>title</th>\n",
+       "      <th>novelty</th>\n",
+       "      <th>date</th>\n",
+       "      <th>in_training</th>\n",
+       "      <th>train/before_mean</th>\n",
+       "      <th>train/after_mean</th>\n",
+       "      <th>train/diff_mean</th>\n",
+       "      <th>train/diff%_mean</th>\n",
+       "      <th>train/before_std</th>\n",
+       "      <th>train/after_std</th>\n",
+       "      <th>...</th>\n",
+       "      <th>train/diff_min</th>\n",
+       "      <th>train/diff%_min</th>\n",
+       "      <th>train/before_max</th>\n",
+       "      <th>train/after_max</th>\n",
+       "      <th>train/diff_max</th>\n",
+       "      <th>train/diff%_max</th>\n",
+       "      <th>train/before_sum</th>\n",
+       "      <th>train/after_sum</th>\n",
+       "      <th>train/diff_sum</th>\n",
+       "      <th>train/diff%_sum</th>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>f</th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
@@ -1073,255 +1919,1406 @@
        "  </thead>\n",
        "  <tbody>\n",
        "    <tr>\n",
-       "      <th>cicero from ibois, Philippe (2012-06-03).</th>\n",
-       "      <td>72.587372</td>\n",
-       "      <td>67.744194</td>\n",
-       "      <td>13707</td>\n",
-       "      <td>0.066722</td>\n",
-       "      <td>4.843178</td>\n",
-       "      <td>True</td>\n",
-       "      <td>True</td>\n",
+       "      <th>../samples/2024_trump_appointment.md</th>\n",
+       "      <td>President Trump Announces Morgan Ortagus as De...</td>\n",
+       "      <td>0.300000</td>\n",
+       "      <td>2025-01-04 00:00:00+00:00</td>\n",
        "      <td>False</td>\n",
+       "      <td>3.602360</td>\n",
+       "      <td>1.976479</td>\n",
+       "      <td>-1.625882</td>\n",
+       "      <td>-45.133784</td>\n",
+       "      <td>3.130189</td>\n",
+       "      <td>2.116151</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.005089</td>\n",
+       "      <td>-82.696441</td>\n",
+       "      <td>17.917255</td>\n",
+       "      <td>11.689822</td>\n",
+       "      <td>-6.227433</td>\n",
+       "      <td>-34.756625</td>\n",
+       "      <td>284.586479</td>\n",
+       "      <td>156.141833</td>\n",
+       "      <td>-128.444646</td>\n",
+       "      <td>-45.133784</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>politics is the mind-killer</th>\n",
-       "      <td>247.551559</td>\n",
-       "      <td>245.826813</td>\n",
-       "      <td>3158</td>\n",
-       "      <td>0.006967</td>\n",
-       "      <td>1.724747</td>\n",
-       "      <td>True</td>\n",
-       "      <td>False</td>\n",
+       "      <th>../samples/2024_how_to_focus.md</th>\n",
+       "      <td>How to Focus</td>\n",
+       "      <td>0.500000</td>\n",
+       "      <td>2024-06-01 00:00:00+00:00</td>\n",
        "      <td>False</td>\n",
+       "      <td>2.030182</td>\n",
+       "      <td>1.238443</td>\n",
+       "      <td>-0.791739</td>\n",
+       "      <td>-38.998412</td>\n",
+       "      <td>2.826893</td>\n",
+       "      <td>2.380078</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.000099</td>\n",
+       "      <td>-40.660140</td>\n",
+       "      <td>13.456800</td>\n",
+       "      <td>13.169463</td>\n",
+       "      <td>-0.287336</td>\n",
+       "      <td>-2.135250</td>\n",
+       "      <td>347.161163</td>\n",
+       "      <td>211.773822</td>\n",
+       "      <td>-135.387340</td>\n",
+       "      <td>-38.998412</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>openai board ann</th>\n",
-       "      <td>55.808502</td>\n",
-       "      <td>54.667923</td>\n",
-       "      <td>2991</td>\n",
-       "      <td>0.020437</td>\n",
-       "      <td>1.140579</td>\n",
-       "      <td>True</td>\n",
-       "      <td>True</td>\n",
+       "      <th>../samples/2024_gpt4_fake_paper.md</th>\n",
+       "      <td>fake ai hoax paper made up by gpt-4</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>2024-01-01 00:00:00+00:00</td>\n",
        "      <td>False</td>\n",
+       "      <td>2.890543</td>\n",
+       "      <td>1.876897</td>\n",
+       "      <td>-1.013646</td>\n",
+       "      <td>-35.067661</td>\n",
+       "      <td>2.491279</td>\n",
+       "      <td>2.017982</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.000804</td>\n",
+       "      <td>-71.474606</td>\n",
+       "      <td>11.427372</td>\n",
+       "      <td>9.812000</td>\n",
+       "      <td>-1.615372</td>\n",
+       "      <td>-14.135986</td>\n",
+       "      <td>349.755680</td>\n",
+       "      <td>227.104543</td>\n",
+       "      <td>-122.651137</td>\n",
+       "      <td>-35.067661</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>How to Catch an AI Liar</th>\n",
-       "      <td>28.949860</td>\n",
-       "      <td>28.008789</td>\n",
-       "      <td>5464</td>\n",
-       "      <td>0.032507</td>\n",
-       "      <td>0.941071</td>\n",
-       "      <td>True</td>\n",
-       "      <td>True</td>\n",
+       "      <th>../samples/2025_h5n1_report.md</th>\n",
+       "      <td>CDC Report on Missouri H5N1 Serology Testing</td>\n",
+       "      <td>0.750000</td>\n",
+       "      <td>2025-01-05 00:00:00+00:00</td>\n",
        "      <td>False</td>\n",
+       "      <td>2.840106</td>\n",
+       "      <td>1.588481</td>\n",
+       "      <td>-1.251626</td>\n",
+       "      <td>-44.069677</td>\n",
+       "      <td>2.737036</td>\n",
+       "      <td>2.004289</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.002226</td>\n",
+       "      <td>-77.666175</td>\n",
+       "      <td>11.928956</td>\n",
+       "      <td>9.404486</td>\n",
+       "      <td>-2.524470</td>\n",
+       "      <td>-21.162542</td>\n",
+       "      <td>355.013296</td>\n",
+       "      <td>198.560084</td>\n",
+       "      <td>-156.453212</td>\n",
+       "      <td>-44.069677</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>buzzfeed foi fauci emails 2023</th>\n",
-       "      <td>23.309443</td>\n",
-       "      <td>22.406410</td>\n",
-       "      <td>13640</td>\n",
-       "      <td>0.038741</td>\n",
-       "      <td>0.903032</td>\n",
-       "      <td>True</td>\n",
-       "      <td>True</td>\n",
+       "      <th>../samples/2024_arxiv_meh.md</th>\n",
+       "      <td>TradingAgents: Multi-Agents LLM Financial Trad...</td>\n",
+       "      <td>0.150000</td>\n",
+       "      <td>2024-12-28 00:00:00+00:00</td>\n",
        "      <td>False</td>\n",
+       "      <td>3.329921</td>\n",
+       "      <td>2.693382</td>\n",
+       "      <td>-0.636539</td>\n",
+       "      <td>-19.115730</td>\n",
+       "      <td>3.075030</td>\n",
+       "      <td>2.940876</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.000330</td>\n",
+       "      <td>-61.776143</td>\n",
+       "      <td>13.523319</td>\n",
+       "      <td>13.253697</td>\n",
+       "      <td>-0.269622</td>\n",
+       "      <td>-1.993755</td>\n",
+       "      <td>386.270819</td>\n",
+       "      <td>312.432330</td>\n",
+       "      <td>-73.838489</td>\n",
+       "      <td>-19.115730</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>Gemini to Q*</th>\n",
-       "      <td>11.756384</td>\n",
-       "      <td>11.190644</td>\n",
-       "      <td>42604</td>\n",
-       "      <td>0.048122</td>\n",
-       "      <td>0.565740</td>\n",
-       "      <td>False</td>\n",
-       "      <td>True</td>\n",
+       "      <th>../samples/2024_lw_by-default-capital-will-matter-more-than-ever-after-agi.md</th>\n",
+       "      <td>By default, capital will matter more than ever...</td>\n",
+       "      <td>0.906388</td>\n",
+       "      <td>2024-12-30 19:47:19.838000+00:00</td>\n",
        "      <td>False</td>\n",
+       "      <td>2.935460</td>\n",
+       "      <td>1.667801</td>\n",
+       "      <td>-1.267660</td>\n",
+       "      <td>-43.184356</td>\n",
+       "      <td>2.497172</td>\n",
+       "      <td>2.004676</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.007733</td>\n",
+       "      <td>-65.756404</td>\n",
+       "      <td>12.706497</td>\n",
+       "      <td>11.517975</td>\n",
+       "      <td>-1.188522</td>\n",
+       "      <td>-9.353658</td>\n",
+       "      <td>393.351672</td>\n",
+       "      <td>223.485284</td>\n",
+       "      <td>-169.866388</td>\n",
+       "      <td>-43.184356</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>LK-99-en</th>\n",
-       "      <td>14.513845</td>\n",
-       "      <td>14.066093</td>\n",
-       "      <td>15432</td>\n",
-       "      <td>0.030850</td>\n",
-       "      <td>0.447752</td>\n",
-       "      <td>False</td>\n",
-       "      <td>True</td>\n",
+       "      <th>../samples/2024_news_anthropic.md</th>\n",
+       "      <td>Amazon-backed Anthropic debuts AI agents that ...</td>\n",
+       "      <td>0.500000</td>\n",
+       "      <td>2025-01-05 05:03:00+00:00</td>\n",
        "      <td>False</td>\n",
+       "      <td>2.437276</td>\n",
+       "      <td>1.420116</td>\n",
+       "      <td>-1.017160</td>\n",
+       "      <td>-41.733471</td>\n",
+       "      <td>2.880203</td>\n",
+       "      <td>2.245643</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.000029</td>\n",
+       "      <td>-71.181132</td>\n",
+       "      <td>12.790111</td>\n",
+       "      <td>12.667040</td>\n",
+       "      <td>-0.123071</td>\n",
+       "      <td>-0.962233</td>\n",
+       "      <td>424.085983</td>\n",
+       "      <td>247.100184</td>\n",
+       "      <td>-176.985800</td>\n",
+       "      <td>-41.733471</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>LK-99-es</th>\n",
-       "      <td>11.415018</td>\n",
-       "      <td>10.972948</td>\n",
-       "      <td>12970</td>\n",
-       "      <td>0.038727</td>\n",
-       "      <td>0.442070</td>\n",
-       "      <td>False</td>\n",
-       "      <td>True</td>\n",
+       "      <th>../samples/2025_lw_the-laws-of-large-numbers.md</th>\n",
+       "      <td>The Laws of Large Numbers</td>\n",
+       "      <td>0.540932</td>\n",
+       "      <td>2025-01-04 18:06:02.387000+00:00</td>\n",
        "      <td>False</td>\n",
+       "      <td>2.887975</td>\n",
+       "      <td>2.153969</td>\n",
+       "      <td>-0.734006</td>\n",
+       "      <td>-25.415931</td>\n",
+       "      <td>2.576897</td>\n",
+       "      <td>2.052082</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.001162</td>\n",
+       "      <td>-29.038092</td>\n",
+       "      <td>11.118690</td>\n",
+       "      <td>8.321499</td>\n",
+       "      <td>-2.797191</td>\n",
+       "      <td>-25.157557</td>\n",
+       "      <td>424.532354</td>\n",
+       "      <td>316.633504</td>\n",
+       "      <td>-107.898850</td>\n",
+       "      <td>-25.415931</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>disney appointment</th>\n",
-       "      <td>118.826157</td>\n",
-       "      <td>118.420128</td>\n",
-       "      <td>3653</td>\n",
-       "      <td>0.003417</td>\n",
-       "      <td>0.406029</td>\n",
+       "      <th>../samples/politics_is_the_mind_killer.md</th>\n",
+       "      <td>politics is the mind-killer</td>\n",
+       "      <td>0.500000</td>\n",
+       "      <td>2007-02-19 00:00:00+00:00</td>\n",
        "      <td>True</td>\n",
-       "      <td>False</td>\n",
-       "      <td>False</td>\n",
+       "      <td>3.285058</td>\n",
+       "      <td>2.430624</td>\n",
+       "      <td>-0.854434</td>\n",
+       "      <td>-26.009715</td>\n",
+       "      <td>2.930715</td>\n",
+       "      <td>2.513867</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.004398</td>\n",
+       "      <td>-65.298689</td>\n",
+       "      <td>11.889731</td>\n",
+       "      <td>11.973468</td>\n",
+       "      <td>0.083736</td>\n",
+       "      <td>0.704275</td>\n",
+       "      <td>446.767846</td>\n",
+       "      <td>330.564804</td>\n",
+       "      <td>-116.203043</td>\n",
+       "      <td>-26.009715</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>weak to strong</th>\n",
-       "      <td>46.764183</td>\n",
-       "      <td>46.404736</td>\n",
-       "      <td>5811</td>\n",
-       "      <td>0.007686</td>\n",
-       "      <td>0.359447</td>\n",
-       "      <td>True</td>\n",
-       "      <td>False</td>\n",
+       "      <th>../samples/2024_lesswrong_slop.md</th>\n",
+       "      <td>Deontic Explorations In \"Paying To Talk To Sla...</td>\n",
+       "      <td>0.100000</td>\n",
+       "      <td>2024-04-12 00:00:00+00:00</td>\n",
        "      <td>False</td>\n",
+       "      <td>3.078750</td>\n",
+       "      <td>1.695459</td>\n",
+       "      <td>-1.383291</td>\n",
+       "      <td>-44.930269</td>\n",
+       "      <td>2.946400</td>\n",
+       "      <td>2.349493</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.000778</td>\n",
+       "      <td>-75.973899</td>\n",
+       "      <td>15.304575</td>\n",
+       "      <td>12.805290</td>\n",
+       "      <td>-2.499285</td>\n",
+       "      <td>-16.330311</td>\n",
+       "      <td>449.497469</td>\n",
+       "      <td>247.537046</td>\n",
+       "      <td>-201.960423</td>\n",
+       "      <td>-44.930269</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>blechley declaration</th>\n",
-       "      <td>17.869091</td>\n",
-       "      <td>17.524204</td>\n",
-       "      <td>7762</td>\n",
-       "      <td>0.019301</td>\n",
-       "      <td>0.344887</td>\n",
-       "      <td>True</td>\n",
-       "      <td>False</td>\n",
+       "      <th>../samples/2025_lw_parkinson-s-law-and-the-ideology-of-statistics-1.md</th>\n",
+       "      <td>Parkinson's Law and the Ideology of Statistics</td>\n",
+       "      <td>0.677458</td>\n",
+       "      <td>2025-01-04 22:59:57.376000+00:00</td>\n",
        "      <td>False</td>\n",
+       "      <td>3.550214</td>\n",
+       "      <td>2.779585</td>\n",
+       "      <td>-0.770629</td>\n",
+       "      <td>-21.706555</td>\n",
+       "      <td>3.111395</td>\n",
+       "      <td>2.862761</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.004117</td>\n",
+       "      <td>-90.026555</td>\n",
+       "      <td>12.598433</td>\n",
+       "      <td>12.334391</td>\n",
+       "      <td>-0.264043</td>\n",
+       "      <td>-2.095839</td>\n",
+       "      <td>450.877187</td>\n",
+       "      <td>353.007281</td>\n",
+       "      <td>-97.869906</td>\n",
+       "      <td>-21.706555</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>Lorem ipsum</th>\n",
-       "      <td>6.564840</td>\n",
-       "      <td>6.268792</td>\n",
-       "      <td>19649</td>\n",
-       "      <td>0.045096</td>\n",
-       "      <td>0.296049</td>\n",
+       "      <th>../samples/2024_anthropic_palintir.md</th>\n",
+       "      <td>Anthropic and Palantir Partner to Bring Claude...</td>\n",
+       "      <td>0.200000</td>\n",
+       "      <td>2024-07-11 00:00:00+00:00</td>\n",
        "      <td>False</td>\n",
-       "      <td>True</td>\n",
-       "      <td>True</td>\n",
+       "      <td>3.575668</td>\n",
+       "      <td>2.444458</td>\n",
+       "      <td>-1.131210</td>\n",
+       "      <td>-31.636323</td>\n",
+       "      <td>3.246451</td>\n",
+       "      <td>2.673633</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.000074</td>\n",
+       "      <td>-75.181568</td>\n",
+       "      <td>13.271792</td>\n",
+       "      <td>12.891874</td>\n",
+       "      <td>-0.379918</td>\n",
+       "      <td>-2.862598</td>\n",
+       "      <td>454.109781</td>\n",
+       "      <td>310.446145</td>\n",
+       "      <td>-143.663636</td>\n",
+       "      <td>-31.636323</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>statement by whitehouse on passing</th>\n",
-       "      <td>29.197132</td>\n",
-       "      <td>28.939692</td>\n",
-       "      <td>1641</td>\n",
-       "      <td>0.008817</td>\n",
-       "      <td>0.257441</td>\n",
-       "      <td>True</td>\n",
+       "      <th>../samples/2025_lw_preference-inversion.md</th>\n",
+       "      <td>Preference Inversion</td>\n",
+       "      <td>0.633321</td>\n",
+       "      <td>2025-01-03 23:49:06.168000+00:00</td>\n",
        "      <td>False</td>\n",
-       "      <td>True</td>\n",
+       "      <td>3.358938</td>\n",
+       "      <td>1.602391</td>\n",
+       "      <td>-1.756548</td>\n",
+       "      <td>-52.294728</td>\n",
+       "      <td>2.745416</td>\n",
+       "      <td>1.724408</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.003750</td>\n",
+       "      <td>-88.162939</td>\n",
+       "      <td>11.404502</td>\n",
+       "      <td>8.079728</td>\n",
+       "      <td>-3.324774</td>\n",
+       "      <td>-29.153170</td>\n",
+       "      <td>456.815627</td>\n",
+       "      <td>217.925137</td>\n",
+       "      <td>-238.890490</td>\n",
+       "      <td>-52.294728</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>harvard announcment caplain israel hamas</th>\n",
-       "      <td>45.347355</td>\n",
-       "      <td>45.127274</td>\n",
-       "      <td>4247</td>\n",
-       "      <td>0.004853</td>\n",
-       "      <td>0.220081</td>\n",
-       "      <td>True</td>\n",
+       "      <th>../samples/2024_deliberative_alignment.md</th>\n",
+       "      <td>Deliberative Alignment: Reasoning Enables Safe...</td>\n",
+       "      <td>0.600000</td>\n",
+       "      <td>2024-12-20 00:00:00+00:00</td>\n",
        "      <td>False</td>\n",
-       "      <td>True</td>\n",
+       "      <td>3.623786</td>\n",
+       "      <td>2.384146</td>\n",
+       "      <td>-1.239640</td>\n",
+       "      <td>-34.208421</td>\n",
+       "      <td>3.441515</td>\n",
+       "      <td>2.720267</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.000561</td>\n",
+       "      <td>-29.432378</td>\n",
+       "      <td>14.758335</td>\n",
+       "      <td>12.259495</td>\n",
+       "      <td>-2.498840</td>\n",
+       "      <td>-16.931722</td>\n",
+       "      <td>467.468432</td>\n",
+       "      <td>307.554861</td>\n",
+       "      <td>-159.913571</td>\n",
+       "      <td>-34.208421</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>fake ai hoax paper</th>\n",
-       "      <td>7.766979</td>\n",
-       "      <td>7.697229</td>\n",
-       "      <td>3290</td>\n",
-       "      <td>0.008980</td>\n",
-       "      <td>0.069750</td>\n",
+       "      <th>../samples/2025_lw_the-subset-parity-learning-problem-much-more-than-you-wanted.md</th>\n",
+       "      <td>The subset parity learning problem: much more ...</td>\n",
+       "      <td>0.697946</td>\n",
+       "      <td>2025-01-04 09:59:16.158000+00:00</td>\n",
        "      <td>False</td>\n",
+       "      <td>3.322798</td>\n",
+       "      <td>1.738719</td>\n",
+       "      <td>-1.584079</td>\n",
+       "      <td>-47.673055</td>\n",
+       "      <td>2.968027</td>\n",
+       "      <td>2.028070</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.005546</td>\n",
+       "      <td>-71.898714</td>\n",
+       "      <td>12.878409</td>\n",
+       "      <td>10.411036</td>\n",
+       "      <td>-2.467374</td>\n",
+       "      <td>-19.158995</td>\n",
+       "      <td>471.837339</td>\n",
+       "      <td>246.898066</td>\n",
+       "      <td>-224.939272</td>\n",
+       "      <td>-47.673055</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2025_lw_human-study-on-ai-spear-phishing-campaigns.md</th>\n",
+       "      <td>Human study on AI spear phishing campaigns</td>\n",
+       "      <td>0.677458</td>\n",
+       "      <td>2025-01-03 19:03:28.406000+00:00</td>\n",
        "      <td>False</td>\n",
+       "      <td>3.436050</td>\n",
+       "      <td>1.997053</td>\n",
+       "      <td>-1.438996</td>\n",
+       "      <td>-41.879381</td>\n",
+       "      <td>3.160585</td>\n",
+       "      <td>2.289065</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.002553</td>\n",
+       "      <td>-85.223138</td>\n",
+       "      <td>14.798123</td>\n",
+       "      <td>12.255587</td>\n",
+       "      <td>-2.542537</td>\n",
+       "      <td>-17.181481</td>\n",
+       "      <td>477.610894</td>\n",
+       "      <td>277.590406</td>\n",
+       "      <td>-200.020488</td>\n",
+       "      <td>-41.879381</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2025_lw_the-intelligence-curse.md</th>\n",
+       "      <td>The Intelligence Curse</td>\n",
+       "      <td>0.688044</td>\n",
+       "      <td>2025-01-04 18:16:58.921000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "      <td>3.539096</td>\n",
+       "      <td>1.598036</td>\n",
+       "      <td>-1.941060</td>\n",
+       "      <td>-54.846207</td>\n",
+       "      <td>3.480139</td>\n",
+       "      <td>2.130229</td>\n",
+       "      <td>...</td>\n",
+       "      <td>0.000008</td>\n",
+       "      <td>2.610249</td>\n",
+       "      <td>17.431973</td>\n",
+       "      <td>11.515382</td>\n",
+       "      <td>-5.916591</td>\n",
+       "      <td>-33.941028</td>\n",
+       "      <td>484.856104</td>\n",
+       "      <td>218.930921</td>\n",
+       "      <td>-265.925183</td>\n",
+       "      <td>-54.846207</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/lorem_ipsum.md</th>\n",
+       "      <td>Lorem ipsum</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>1900-01-01 00:00:00+00:00</td>\n",
        "      <td>True</td>\n",
+       "      <td>2.465992</td>\n",
+       "      <td>1.825440</td>\n",
+       "      <td>-0.640552</td>\n",
+       "      <td>-25.975428</td>\n",
+       "      <td>2.463530</td>\n",
+       "      <td>2.008885</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.000011</td>\n",
+       "      <td>-19.999573</td>\n",
+       "      <td>12.736030</td>\n",
+       "      <td>12.230991</td>\n",
+       "      <td>-0.505038</td>\n",
+       "      <td>-3.965429</td>\n",
+       "      <td>490.732376</td>\n",
+       "      <td>363.262540</td>\n",
+       "      <td>-127.469837</td>\n",
+       "      <td>-25.975428</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2025_lw_the-field-of-ai-alignment-a-postmortem-and-what-to-do-about.md</th>\n",
+       "      <td>The Field of AI Alignment: A Postmortem, and W...</td>\n",
+       "      <td>0.925483</td>\n",
+       "      <td>2025-01-01 00:42:36.538000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "      <td>3.735233</td>\n",
+       "      <td>1.579777</td>\n",
+       "      <td>-2.155455</td>\n",
+       "      <td>-57.706054</td>\n",
+       "      <td>3.116857</td>\n",
+       "      <td>2.056279</td>\n",
+       "      <td>...</td>\n",
+       "      <td>0.000621</td>\n",
+       "      <td>378.210283</td>\n",
+       "      <td>12.190395</td>\n",
+       "      <td>10.575907</td>\n",
+       "      <td>-1.614489</td>\n",
+       "      <td>-13.243940</td>\n",
+       "      <td>500.521191</td>\n",
+       "      <td>211.690162</td>\n",
+       "      <td>-288.831029</td>\n",
+       "      <td>-57.706054</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2024_openai_emails.md</th>\n",
+       "      <td>OpenAI Email Archives from Musk v. Altman</td>\n",
+       "      <td>0.700000</td>\n",
+       "      <td>2024-11-01 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "      <td>3.426089</td>\n",
+       "      <td>2.140106</td>\n",
+       "      <td>-1.285983</td>\n",
+       "      <td>-37.535013</td>\n",
+       "      <td>2.755313</td>\n",
+       "      <td>2.184135</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.000420</td>\n",
+       "      <td>-95.082707</td>\n",
+       "      <td>10.482502</td>\n",
+       "      <td>9.602067</td>\n",
+       "      <td>-0.880435</td>\n",
+       "      <td>-8.399092</td>\n",
+       "      <td>517.339424</td>\n",
+       "      <td>323.156004</td>\n",
+       "      <td>-194.183420</td>\n",
+       "      <td>-37.535013</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2025_lw_debating-buying-nvda-in-2019.md</th>\n",
+       "      <td>debating buying NVDA in 2019</td>\n",
+       "      <td>0.526975</td>\n",
+       "      <td>2025-01-04 18:07:35.832000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "      <td>3.498877</td>\n",
+       "      <td>1.946026</td>\n",
+       "      <td>-1.552851</td>\n",
+       "      <td>-44.381423</td>\n",
+       "      <td>2.970988</td>\n",
+       "      <td>2.091860</td>\n",
+       "      <td>...</td>\n",
+       "      <td>0.000716</td>\n",
+       "      <td>144.210285</td>\n",
+       "      <td>12.902117</td>\n",
+       "      <td>9.158434</td>\n",
+       "      <td>-3.743683</td>\n",
+       "      <td>-29.016036</td>\n",
+       "      <td>517.833801</td>\n",
+       "      <td>288.011791</td>\n",
+       "      <td>-229.822011</td>\n",
+       "      <td>-44.381423</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2025_lw_my-agi-safety-research-2024-review-25-plans.md</th>\n",
+       "      <td>My AGI safety research—2024 review, ’25 plans</td>\n",
+       "      <td>0.756925</td>\n",
+       "      <td>2025-01-01 22:01:48.820000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "      <td>3.153156</td>\n",
+       "      <td>1.580539</td>\n",
+       "      <td>-1.572617</td>\n",
+       "      <td>-49.874378</td>\n",
+       "      <td>3.112932</td>\n",
+       "      <td>2.097326</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.001612</td>\n",
+       "      <td>-86.930632</td>\n",
+       "      <td>13.688265</td>\n",
+       "      <td>11.766024</td>\n",
+       "      <td>-1.922241</td>\n",
+       "      <td>-14.042987</td>\n",
+       "      <td>523.423819</td>\n",
+       "      <td>262.369446</td>\n",
+       "      <td>-261.054373</td>\n",
+       "      <td>-49.874378</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2024_bob_fanfic.md</th>\n",
+       "      <td>Flower Crowns and Furry Mishaps by MyPalAI</td>\n",
+       "      <td>0.300000</td>\n",
+       "      <td>2024-12-24 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "      <td>3.314768</td>\n",
+       "      <td>2.646716</td>\n",
+       "      <td>-0.668052</td>\n",
+       "      <td>-20.153819</td>\n",
+       "      <td>3.029083</td>\n",
+       "      <td>2.704339</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.000615</td>\n",
+       "      <td>-35.032627</td>\n",
+       "      <td>13.226252</td>\n",
+       "      <td>11.086496</td>\n",
+       "      <td>-2.139755</td>\n",
+       "      <td>-16.178093</td>\n",
+       "      <td>527.048090</td>\n",
+       "      <td>420.827773</td>\n",
+       "      <td>-106.220317</td>\n",
+       "      <td>-20.153819</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2024_gwern_reddit.md</th>\n",
+       "      <td>Hardware Hedging Against Scaling Regime Shifts...</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>2024-08-21 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "      <td>3.907376</td>\n",
+       "      <td>2.918248</td>\n",
+       "      <td>-0.989128</td>\n",
+       "      <td>-25.314383</td>\n",
+       "      <td>3.043581</td>\n",
+       "      <td>2.630350</td>\n",
+       "      <td>...</td>\n",
+       "      <td>0.000549</td>\n",
+       "      <td>46.703333</td>\n",
+       "      <td>15.920402</td>\n",
+       "      <td>14.091064</td>\n",
+       "      <td>-1.829337</td>\n",
+       "      <td>-11.490521</td>\n",
+       "      <td>562.662201</td>\n",
+       "      <td>420.227737</td>\n",
+       "      <td>-142.434464</td>\n",
+       "      <td>-25.314383</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2025_lw_comment-on-death-and-the-gorgon.md</th>\n",
+       "      <td>Comment on 'Death and the Gorgon'</td>\n",
+       "      <td>0.750253</td>\n",
+       "      <td>2025-01-01 06:00:24.498000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "      <td>4.084834</td>\n",
+       "      <td>2.495894</td>\n",
+       "      <td>-1.588940</td>\n",
+       "      <td>-38.898521</td>\n",
+       "      <td>3.372985</td>\n",
+       "      <td>2.744499</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.001318</td>\n",
+       "      <td>-65.735003</td>\n",
+       "      <td>16.383120</td>\n",
+       "      <td>15.471860</td>\n",
+       "      <td>-0.911260</td>\n",
+       "      <td>-5.562186</td>\n",
+       "      <td>616.809882</td>\n",
+       "      <td>376.879962</td>\n",
+       "      <td>-239.929921</td>\n",
+       "      <td>-38.898521</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2024_bob_fanfic2.md</th>\n",
+       "      <td>Paradox's Box (Bobiverse) by Mark4man</td>\n",
+       "      <td>0.400000</td>\n",
+       "      <td>2022-02-17 00:00:00+00:00</td>\n",
+       "      <td>True</td>\n",
+       "      <td>4.136299</td>\n",
+       "      <td>3.299236</td>\n",
+       "      <td>-0.837063</td>\n",
+       "      <td>-20.237010</td>\n",
+       "      <td>3.103517</td>\n",
+       "      <td>2.757547</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.001771</td>\n",
+       "      <td>-35.546726</td>\n",
+       "      <td>13.576470</td>\n",
+       "      <td>12.110069</td>\n",
+       "      <td>-1.466401</td>\n",
+       "      <td>-10.801048</td>\n",
+       "      <td>620.444836</td>\n",
+       "      <td>494.885355</td>\n",
+       "      <td>-125.559481</td>\n",
+       "      <td>-20.237010</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2025_lw_what-s-the-short-timeline-plan.md</th>\n",
+       "      <td>What’s the short timeline plan?</td>\n",
+       "      <td>0.898161</td>\n",
+       "      <td>2025-01-05 00:10:28.708000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "      <td>3.970715</td>\n",
+       "      <td>3.319989</td>\n",
+       "      <td>-0.650725</td>\n",
+       "      <td>-16.388113</td>\n",
+       "      <td>3.171675</td>\n",
+       "      <td>2.998089</td>\n",
+       "      <td>...</td>\n",
+       "      <td>0.000231</td>\n",
+       "      <td>159.805322</td>\n",
+       "      <td>13.405236</td>\n",
+       "      <td>17.062244</td>\n",
+       "      <td>3.657008</td>\n",
+       "      <td>27.280446</td>\n",
+       "      <td>623.402177</td>\n",
+       "      <td>521.238323</td>\n",
+       "      <td>-102.163854</td>\n",
+       "      <td>-16.388113</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2024_lw_the-plan-2024-update.md</th>\n",
+       "      <td>The Plan - 2024 Update</td>\n",
+       "      <td>0.785170</td>\n",
+       "      <td>2024-12-31 19:29:23.013000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "      <td>3.741762</td>\n",
+       "      <td>2.981632</td>\n",
+       "      <td>-0.760130</td>\n",
+       "      <td>-20.314768</td>\n",
+       "      <td>2.872466</td>\n",
+       "      <td>2.731928</td>\n",
+       "      <td>...</td>\n",
+       "      <td>0.000043</td>\n",
+       "      <td>70.526800</td>\n",
+       "      <td>11.613052</td>\n",
+       "      <td>11.740885</td>\n",
+       "      <td>0.127832</td>\n",
+       "      <td>1.100765</td>\n",
+       "      <td>673.517207</td>\n",
+       "      <td>536.693749</td>\n",
+       "      <td>-136.823458</td>\n",
+       "      <td>-20.314768</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2025_lw_review-planecrash.md</th>\n",
+       "      <td>Review: Planecrash</td>\n",
+       "      <td>0.933950</td>\n",
+       "      <td>2025-01-01 01:31:26.864000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "      <td>3.475632</td>\n",
+       "      <td>1.752375</td>\n",
+       "      <td>-1.723257</td>\n",
+       "      <td>-49.581118</td>\n",
+       "      <td>3.319073</td>\n",
+       "      <td>2.316245</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-0.000083</td>\n",
+       "      <td>-79.884876</td>\n",
+       "      <td>14.514912</td>\n",
+       "      <td>10.882781</td>\n",
+       "      <td>-3.632131</td>\n",
+       "      <td>-25.023443</td>\n",
+       "      <td>684.699408</td>\n",
+       "      <td>345.217788</td>\n",
+       "      <td>-339.481621</td>\n",
+       "      <td>-49.581118</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>../samples/2025_lw_2024-in-ai-predictions.md</th>\n",
+       "      <td>2024 in AI predictions</td>\n",
+       "      <td>0.789190</td>\n",
+       "      <td>2025-01-02 02:09:13.093000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "      <td>4.635499</td>\n",
+       "      <td>2.916743</td>\n",
+       "      <td>-1.718757</td>\n",
+       "      <td>-37.078135</td>\n",
+       "      <td>3.080130</td>\n",
+       "      <td>2.548945</td>\n",
+       "      <td>...</td>\n",
+       "      <td>0.000238</td>\n",
+       "      <td>50.615295</td>\n",
+       "      <td>13.301698</td>\n",
+       "      <td>12.898935</td>\n",
+       "      <td>-0.402762</td>\n",
+       "      <td>-3.027902</td>\n",
+       "      <td>801.941370</td>\n",
+       "      <td>504.596465</td>\n",
+       "      <td>-297.344904</td>\n",
+       "      <td>-37.078135</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
+       "<p>30 rows × 24 columns</p>\n",
        "</div>"
       ],
       "text/plain": [
-       "                                               before       after    len  \\\n",
-       "title                                                                      \n",
-       "cicero from ibois, Philippe (2012-06-03).   72.587372   67.744194  13707   \n",
-       "politics is the mind-killer                247.551559  245.826813   3158   \n",
-       "openai board ann                            55.808502   54.667923   2991   \n",
-       "How to Catch an AI Liar                     28.949860   28.008789   5464   \n",
-       "buzzfeed foi fauci emails 2023              23.309443   22.406410  13640   \n",
-       "Gemini to Q*                                11.756384   11.190644  42604   \n",
-       "LK-99-en                                    14.513845   14.066093  15432   \n",
-       "LK-99-es                                    11.415018   10.972948  12970   \n",
-       "disney appointment                         118.826157  118.420128   3653   \n",
-       "weak to strong                              46.764183   46.404736   5811   \n",
-       "blechley declaration                        17.869091   17.524204   7762   \n",
-       "Lorem ipsum                                  6.564840    6.268792  19649   \n",
-       "statement by whitehouse on passing          29.197132   28.939692   1641   \n",
-       "harvard announcment caplain israel hamas    45.347355   45.127274   4247   \n",
-       "fake ai hoax paper                           7.766979    7.697229   3290   \n",
+       "                                                                                                title  \\\n",
+       "f                                                                                                       \n",
+       "../samples/2024_trump_appointment.md                President Trump Announces Morgan Ortagus as De...   \n",
+       "../samples/2024_how_to_focus.md                                                          How to Focus   \n",
+       "../samples/2024_gpt4_fake_paper.md                                fake ai hoax paper made up by gpt-4   \n",
+       "../samples/2025_h5n1_report.md                           CDC Report on Missouri H5N1 Serology Testing   \n",
+       "../samples/2024_arxiv_meh.md                        TradingAgents: Multi-Agents LLM Financial Trad...   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...  By default, capital will matter more than ever...   \n",
+       "../samples/2024_news_anthropic.md                   Amazon-backed Anthropic debuts AI agents that ...   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md                             The Laws of Large Numbers   \n",
+       "../samples/politics_is_the_mind_killer.md                                 politics is the mind-killer   \n",
+       "../samples/2024_lesswrong_slop.md                   Deontic Explorations In \"Paying To Talk To Sla...   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...     Parkinson's Law and the Ideology of Statistics   \n",
+       "../samples/2024_anthropic_palintir.md               Anthropic and Palantir Partner to Bring Claude...   \n",
+       "../samples/2025_lw_preference-inversion.md                                       Preference Inversion   \n",
+       "../samples/2024_deliberative_alignment.md           Deliberative Alignment: Reasoning Enables Safe...   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...  The subset parity learning problem: much more ...   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...         Human study on AI spear phishing campaigns   \n",
+       "../samples/2025_lw_the-intelligence-curse.md                                   The Intelligence Curse   \n",
+       "../samples/lorem_ipsum.md                                                                 Lorem ipsum   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...  The Field of AI Alignment: A Postmortem, and W...   \n",
+       "../samples/2024_openai_emails.md                            OpenAI Email Archives from Musk v. Altman   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md                       debating buying NVDA in 2019   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...      My AGI safety research—2024 review, ’25 plans   \n",
+       "../samples/2024_bob_fanfic.md                              Flower Crowns and Furry Mishaps by MyPalAI   \n",
+       "../samples/2024_gwern_reddit.md                     Hardware Hedging Against Scaling Regime Shifts...   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...                  Comment on 'Death and the Gorgon'   \n",
+       "../samples/2024_bob_fanfic2.md                                  Paradox's Box (Bobiverse) by Mark4man   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...                    What’s the short timeline plan?   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                                     The Plan - 2024 Update   \n",
+       "../samples/2025_lw_review-planecrash.md                                            Review: Planecrash   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md                                   2024 in AI predictions   \n",
        "\n",
-       "                                           improvement%  improvement  novel  \\\n",
-       "title                                                                         \n",
-       "cicero from ibois, Philippe (2012-06-03).      0.066722     4.843178   True   \n",
-       "politics is the mind-killer                    0.006967     1.724747   True   \n",
-       "openai board ann                               0.020437     1.140579   True   \n",
-       "How to Catch an AI Liar                        0.032507     0.941071   True   \n",
-       "buzzfeed foi fauci emails 2023                 0.038741     0.903032   True   \n",
-       "Gemini to Q*                                   0.048122     0.565740  False   \n",
-       "LK-99-en                                       0.030850     0.447752  False   \n",
-       "LK-99-es                                       0.038727     0.442070  False   \n",
-       "disney appointment                             0.003417     0.406029   True   \n",
-       "weak to strong                                 0.007686     0.359447   True   \n",
-       "blechley declaration                           0.019301     0.344887   True   \n",
-       "Lorem ipsum                                    0.045096     0.296049  False   \n",
-       "statement by whitehouse on passing             0.008817     0.257441   True   \n",
-       "harvard announcment caplain israel hamas       0.004853     0.220081   True   \n",
-       "fake ai hoax paper                             0.008980     0.069750  False   \n",
+       "                                                     novelty  \\\n",
+       "f                                                              \n",
+       "../samples/2024_trump_appointment.md                0.300000   \n",
+       "../samples/2024_how_to_focus.md                     0.500000   \n",
+       "../samples/2024_gpt4_fake_paper.md                  0.000000   \n",
+       "../samples/2025_h5n1_report.md                      0.750000   \n",
+       "../samples/2024_arxiv_meh.md                        0.150000   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...  0.906388   \n",
+       "../samples/2024_news_anthropic.md                   0.500000   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md     0.540932   \n",
+       "../samples/politics_is_the_mind_killer.md           0.500000   \n",
+       "../samples/2024_lesswrong_slop.md                   0.100000   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...  0.677458   \n",
+       "../samples/2024_anthropic_palintir.md               0.200000   \n",
+       "../samples/2025_lw_preference-inversion.md          0.633321   \n",
+       "../samples/2024_deliberative_alignment.md           0.600000   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...  0.697946   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...  0.677458   \n",
+       "../samples/2025_lw_the-intelligence-curse.md        0.688044   \n",
+       "../samples/lorem_ipsum.md                           0.000000   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...  0.925483   \n",
+       "../samples/2024_openai_emails.md                    0.700000   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md  0.526975   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...  0.756925   \n",
+       "../samples/2024_bob_fanfic.md                       0.300000   \n",
+       "../samples/2024_gwern_reddit.md                     1.000000   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...  0.750253   \n",
+       "../samples/2024_bob_fanfic2.md                      0.400000   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...  0.898161   \n",
+       "../samples/2024_lw_the-plan-2024-update.md          0.785170   \n",
+       "../samples/2025_lw_review-planecrash.md             0.933950   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md        0.789190   \n",
        "\n",
-       "                                           learnable     BS  \n",
-       "title                                                        \n",
-       "cicero from ibois, Philippe (2012-06-03).       True  False  \n",
-       "politics is the mind-killer                    False  False  \n",
-       "openai board ann                                True  False  \n",
-       "How to Catch an AI Liar                         True  False  \n",
-       "buzzfeed foi fauci emails 2023                  True  False  \n",
-       "Gemini to Q*                                    True  False  \n",
-       "LK-99-en                                        True  False  \n",
-       "LK-99-es                                        True  False  \n",
-       "disney appointment                             False  False  \n",
-       "weak to strong                                 False  False  \n",
-       "blechley declaration                           False  False  \n",
-       "Lorem ipsum                                     True   True  \n",
-       "statement by whitehouse on passing             False   True  \n",
-       "harvard announcment caplain israel hamas       False   True  \n",
-       "fake ai hoax paper                             False   True  "
+       "                                                                               date  \\\n",
+       "f                                                                                     \n",
+       "../samples/2024_trump_appointment.md                      2025-01-04 00:00:00+00:00   \n",
+       "../samples/2024_how_to_focus.md                           2024-06-01 00:00:00+00:00   \n",
+       "../samples/2024_gpt4_fake_paper.md                        2024-01-01 00:00:00+00:00   \n",
+       "../samples/2025_h5n1_report.md                            2025-01-05 00:00:00+00:00   \n",
+       "../samples/2024_arxiv_meh.md                              2024-12-28 00:00:00+00:00   \n",
+       "../samples/2024_lw_by-default-capital-will-matt... 2024-12-30 19:47:19.838000+00:00   \n",
+       "../samples/2024_news_anthropic.md                         2025-01-05 05:03:00+00:00   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md    2025-01-04 18:06:02.387000+00:00   \n",
+       "../samples/politics_is_the_mind_killer.md                 2007-02-19 00:00:00+00:00   \n",
+       "../samples/2024_lesswrong_slop.md                         2024-04-12 00:00:00+00:00   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo... 2025-01-04 22:59:57.376000+00:00   \n",
+       "../samples/2024_anthropic_palintir.md                     2024-07-11 00:00:00+00:00   \n",
+       "../samples/2025_lw_preference-inversion.md         2025-01-03 23:49:06.168000+00:00   \n",
+       "../samples/2024_deliberative_alignment.md                 2024-12-20 00:00:00+00:00   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p... 2025-01-04 09:59:16.158000+00:00   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis... 2025-01-03 19:03:28.406000+00:00   \n",
+       "../samples/2025_lw_the-intelligence-curse.md       2025-01-04 18:16:58.921000+00:00   \n",
+       "../samples/lorem_ipsum.md                                 1900-01-01 00:00:00+00:00   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-... 2025-01-01 00:42:36.538000+00:00   \n",
+       "../samples/2024_openai_emails.md                          2024-11-01 00:00:00+00:00   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md 2025-01-04 18:07:35.832000+00:00   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-... 2025-01-01 22:01:48.820000+00:00   \n",
+       "../samples/2024_bob_fanfic.md                             2024-12-24 00:00:00+00:00   \n",
+       "../samples/2024_gwern_reddit.md                           2024-08-21 00:00:00+00:00   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor... 2025-01-01 06:00:24.498000+00:00   \n",
+       "../samples/2024_bob_fanfic2.md                            2022-02-17 00:00:00+00:00   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl... 2025-01-05 00:10:28.708000+00:00   \n",
+       "../samples/2024_lw_the-plan-2024-update.md         2024-12-31 19:29:23.013000+00:00   \n",
+       "../samples/2025_lw_review-planecrash.md            2025-01-01 01:31:26.864000+00:00   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md       2025-01-02 02:09:13.093000+00:00   \n",
+       "\n",
+       "                                                    in_training  \\\n",
+       "f                                                                 \n",
+       "../samples/2024_trump_appointment.md                      False   \n",
+       "../samples/2024_how_to_focus.md                           False   \n",
+       "../samples/2024_gpt4_fake_paper.md                        False   \n",
+       "../samples/2025_h5n1_report.md                            False   \n",
+       "../samples/2024_arxiv_meh.md                              False   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...        False   \n",
+       "../samples/2024_news_anthropic.md                         False   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md           False   \n",
+       "../samples/politics_is_the_mind_killer.md                  True   \n",
+       "../samples/2024_lesswrong_slop.md                         False   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...        False   \n",
+       "../samples/2024_anthropic_palintir.md                     False   \n",
+       "../samples/2025_lw_preference-inversion.md                False   \n",
+       "../samples/2024_deliberative_alignment.md                 False   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...        False   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...        False   \n",
+       "../samples/2025_lw_the-intelligence-curse.md              False   \n",
+       "../samples/lorem_ipsum.md                                  True   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...        False   \n",
+       "../samples/2024_openai_emails.md                          False   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md        False   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...        False   \n",
+       "../samples/2024_bob_fanfic.md                             False   \n",
+       "../samples/2024_gwern_reddit.md                           False   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...        False   \n",
+       "../samples/2024_bob_fanfic2.md                             True   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...        False   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                False   \n",
+       "../samples/2025_lw_review-planecrash.md                   False   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md              False   \n",
+       "\n",
+       "                                                    train/before_mean  \\\n",
+       "f                                                                       \n",
+       "../samples/2024_trump_appointment.md                         3.602360   \n",
+       "../samples/2024_how_to_focus.md                              2.030182   \n",
+       "../samples/2024_gpt4_fake_paper.md                           2.890543   \n",
+       "../samples/2025_h5n1_report.md                               2.840106   \n",
+       "../samples/2024_arxiv_meh.md                                 3.329921   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...           2.935460   \n",
+       "../samples/2024_news_anthropic.md                            2.437276   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md              2.887975   \n",
+       "../samples/politics_is_the_mind_killer.md                    3.285058   \n",
+       "../samples/2024_lesswrong_slop.md                            3.078750   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...           3.550214   \n",
+       "../samples/2024_anthropic_palintir.md                        3.575668   \n",
+       "../samples/2025_lw_preference-inversion.md                   3.358938   \n",
+       "../samples/2024_deliberative_alignment.md                    3.623786   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...           3.322798   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...           3.436050   \n",
+       "../samples/2025_lw_the-intelligence-curse.md                 3.539096   \n",
+       "../samples/lorem_ipsum.md                                    2.465992   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...           3.735233   \n",
+       "../samples/2024_openai_emails.md                             3.426089   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md           3.498877   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...           3.153156   \n",
+       "../samples/2024_bob_fanfic.md                                3.314768   \n",
+       "../samples/2024_gwern_reddit.md                              3.907376   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...           4.084834   \n",
+       "../samples/2024_bob_fanfic2.md                               4.136299   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...           3.970715   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                   3.741762   \n",
+       "../samples/2025_lw_review-planecrash.md                      3.475632   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md                 4.635499   \n",
+       "\n",
+       "                                                    train/after_mean  \\\n",
+       "f                                                                      \n",
+       "../samples/2024_trump_appointment.md                        1.976479   \n",
+       "../samples/2024_how_to_focus.md                             1.238443   \n",
+       "../samples/2024_gpt4_fake_paper.md                          1.876897   \n",
+       "../samples/2025_h5n1_report.md                              1.588481   \n",
+       "../samples/2024_arxiv_meh.md                                2.693382   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...          1.667801   \n",
+       "../samples/2024_news_anthropic.md                           1.420116   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md             2.153969   \n",
+       "../samples/politics_is_the_mind_killer.md                   2.430624   \n",
+       "../samples/2024_lesswrong_slop.md                           1.695459   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...          2.779585   \n",
+       "../samples/2024_anthropic_palintir.md                       2.444458   \n",
+       "../samples/2025_lw_preference-inversion.md                  1.602391   \n",
+       "../samples/2024_deliberative_alignment.md                   2.384146   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...          1.738719   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...          1.997053   \n",
+       "../samples/2025_lw_the-intelligence-curse.md                1.598036   \n",
+       "../samples/lorem_ipsum.md                                   1.825440   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...          1.579777   \n",
+       "../samples/2024_openai_emails.md                            2.140106   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md          1.946026   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...          1.580539   \n",
+       "../samples/2024_bob_fanfic.md                               2.646716   \n",
+       "../samples/2024_gwern_reddit.md                             2.918248   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...          2.495894   \n",
+       "../samples/2024_bob_fanfic2.md                              3.299236   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...          3.319989   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                  2.981632   \n",
+       "../samples/2025_lw_review-planecrash.md                     1.752375   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md                2.916743   \n",
+       "\n",
+       "                                                    train/diff_mean  \\\n",
+       "f                                                                     \n",
+       "../samples/2024_trump_appointment.md                      -1.625882   \n",
+       "../samples/2024_how_to_focus.md                           -0.791739   \n",
+       "../samples/2024_gpt4_fake_paper.md                        -1.013646   \n",
+       "../samples/2025_h5n1_report.md                            -1.251626   \n",
+       "../samples/2024_arxiv_meh.md                              -0.636539   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...        -1.267660   \n",
+       "../samples/2024_news_anthropic.md                         -1.017160   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md           -0.734006   \n",
+       "../samples/politics_is_the_mind_killer.md                 -0.854434   \n",
+       "../samples/2024_lesswrong_slop.md                         -1.383291   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...        -0.770629   \n",
+       "../samples/2024_anthropic_palintir.md                     -1.131210   \n",
+       "../samples/2025_lw_preference-inversion.md                -1.756548   \n",
+       "../samples/2024_deliberative_alignment.md                 -1.239640   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...        -1.584079   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...        -1.438996   \n",
+       "../samples/2025_lw_the-intelligence-curse.md              -1.941060   \n",
+       "../samples/lorem_ipsum.md                                 -0.640552   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...        -2.155455   \n",
+       "../samples/2024_openai_emails.md                          -1.285983   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md        -1.552851   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...        -1.572617   \n",
+       "../samples/2024_bob_fanfic.md                             -0.668052   \n",
+       "../samples/2024_gwern_reddit.md                           -0.989128   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...        -1.588940   \n",
+       "../samples/2024_bob_fanfic2.md                            -0.837063   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...        -0.650725   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                -0.760130   \n",
+       "../samples/2025_lw_review-planecrash.md                   -1.723257   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md              -1.718757   \n",
+       "\n",
+       "                                                    train/diff%_mean  \\\n",
+       "f                                                                      \n",
+       "../samples/2024_trump_appointment.md                      -45.133784   \n",
+       "../samples/2024_how_to_focus.md                           -38.998412   \n",
+       "../samples/2024_gpt4_fake_paper.md                        -35.067661   \n",
+       "../samples/2025_h5n1_report.md                            -44.069677   \n",
+       "../samples/2024_arxiv_meh.md                              -19.115730   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...        -43.184356   \n",
+       "../samples/2024_news_anthropic.md                         -41.733471   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md           -25.415931   \n",
+       "../samples/politics_is_the_mind_killer.md                 -26.009715   \n",
+       "../samples/2024_lesswrong_slop.md                         -44.930269   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...        -21.706555   \n",
+       "../samples/2024_anthropic_palintir.md                     -31.636323   \n",
+       "../samples/2025_lw_preference-inversion.md                -52.294728   \n",
+       "../samples/2024_deliberative_alignment.md                 -34.208421   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...        -47.673055   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...        -41.879381   \n",
+       "../samples/2025_lw_the-intelligence-curse.md              -54.846207   \n",
+       "../samples/lorem_ipsum.md                                 -25.975428   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...        -57.706054   \n",
+       "../samples/2024_openai_emails.md                          -37.535013   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md        -44.381423   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...        -49.874378   \n",
+       "../samples/2024_bob_fanfic.md                             -20.153819   \n",
+       "../samples/2024_gwern_reddit.md                           -25.314383   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...        -38.898521   \n",
+       "../samples/2024_bob_fanfic2.md                            -20.237010   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...        -16.388113   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                -20.314768   \n",
+       "../samples/2025_lw_review-planecrash.md                   -49.581118   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md              -37.078135   \n",
+       "\n",
+       "                                                    train/before_std  \\\n",
+       "f                                                                      \n",
+       "../samples/2024_trump_appointment.md                        3.130189   \n",
+       "../samples/2024_how_to_focus.md                             2.826893   \n",
+       "../samples/2024_gpt4_fake_paper.md                          2.491279   \n",
+       "../samples/2025_h5n1_report.md                              2.737036   \n",
+       "../samples/2024_arxiv_meh.md                                3.075030   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...          2.497172   \n",
+       "../samples/2024_news_anthropic.md                           2.880203   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md             2.576897   \n",
+       "../samples/politics_is_the_mind_killer.md                   2.930715   \n",
+       "../samples/2024_lesswrong_slop.md                           2.946400   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...          3.111395   \n",
+       "../samples/2024_anthropic_palintir.md                       3.246451   \n",
+       "../samples/2025_lw_preference-inversion.md                  2.745416   \n",
+       "../samples/2024_deliberative_alignment.md                   3.441515   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...          2.968027   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...          3.160585   \n",
+       "../samples/2025_lw_the-intelligence-curse.md                3.480139   \n",
+       "../samples/lorem_ipsum.md                                   2.463530   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...          3.116857   \n",
+       "../samples/2024_openai_emails.md                            2.755313   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md          2.970988   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...          3.112932   \n",
+       "../samples/2024_bob_fanfic.md                               3.029083   \n",
+       "../samples/2024_gwern_reddit.md                             3.043581   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...          3.372985   \n",
+       "../samples/2024_bob_fanfic2.md                              3.103517   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...          3.171675   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                  2.872466   \n",
+       "../samples/2025_lw_review-planecrash.md                     3.319073   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md                3.080130   \n",
+       "\n",
+       "                                                    train/after_std  ...  \\\n",
+       "f                                                                    ...   \n",
+       "../samples/2024_trump_appointment.md                       2.116151  ...   \n",
+       "../samples/2024_how_to_focus.md                            2.380078  ...   \n",
+       "../samples/2024_gpt4_fake_paper.md                         2.017982  ...   \n",
+       "../samples/2025_h5n1_report.md                             2.004289  ...   \n",
+       "../samples/2024_arxiv_meh.md                               2.940876  ...   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...         2.004676  ...   \n",
+       "../samples/2024_news_anthropic.md                          2.245643  ...   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md            2.052082  ...   \n",
+       "../samples/politics_is_the_mind_killer.md                  2.513867  ...   \n",
+       "../samples/2024_lesswrong_slop.md                          2.349493  ...   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...         2.862761  ...   \n",
+       "../samples/2024_anthropic_palintir.md                      2.673633  ...   \n",
+       "../samples/2025_lw_preference-inversion.md                 1.724408  ...   \n",
+       "../samples/2024_deliberative_alignment.md                  2.720267  ...   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...         2.028070  ...   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...         2.289065  ...   \n",
+       "../samples/2025_lw_the-intelligence-curse.md               2.130229  ...   \n",
+       "../samples/lorem_ipsum.md                                  2.008885  ...   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...         2.056279  ...   \n",
+       "../samples/2024_openai_emails.md                           2.184135  ...   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md         2.091860  ...   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...         2.097326  ...   \n",
+       "../samples/2024_bob_fanfic.md                              2.704339  ...   \n",
+       "../samples/2024_gwern_reddit.md                            2.630350  ...   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...         2.744499  ...   \n",
+       "../samples/2024_bob_fanfic2.md                             2.757547  ...   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...         2.998089  ...   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                 2.731928  ...   \n",
+       "../samples/2025_lw_review-planecrash.md                    2.316245  ...   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md               2.548945  ...   \n",
+       "\n",
+       "                                                    train/diff_min  \\\n",
+       "f                                                                    \n",
+       "../samples/2024_trump_appointment.md                     -0.005089   \n",
+       "../samples/2024_how_to_focus.md                          -0.000099   \n",
+       "../samples/2024_gpt4_fake_paper.md                       -0.000804   \n",
+       "../samples/2025_h5n1_report.md                           -0.002226   \n",
+       "../samples/2024_arxiv_meh.md                             -0.000330   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...       -0.007733   \n",
+       "../samples/2024_news_anthropic.md                        -0.000029   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md          -0.001162   \n",
+       "../samples/politics_is_the_mind_killer.md                -0.004398   \n",
+       "../samples/2024_lesswrong_slop.md                        -0.000778   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...       -0.004117   \n",
+       "../samples/2024_anthropic_palintir.md                    -0.000074   \n",
+       "../samples/2025_lw_preference-inversion.md               -0.003750   \n",
+       "../samples/2024_deliberative_alignment.md                -0.000561   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...       -0.005546   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...       -0.002553   \n",
+       "../samples/2025_lw_the-intelligence-curse.md              0.000008   \n",
+       "../samples/lorem_ipsum.md                                -0.000011   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...        0.000621   \n",
+       "../samples/2024_openai_emails.md                         -0.000420   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md        0.000716   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...       -0.001612   \n",
+       "../samples/2024_bob_fanfic.md                            -0.000615   \n",
+       "../samples/2024_gwern_reddit.md                           0.000549   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...       -0.001318   \n",
+       "../samples/2024_bob_fanfic2.md                           -0.001771   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...        0.000231   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                0.000043   \n",
+       "../samples/2025_lw_review-planecrash.md                  -0.000083   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md              0.000238   \n",
+       "\n",
+       "                                                    train/diff%_min  \\\n",
+       "f                                                                     \n",
+       "../samples/2024_trump_appointment.md                     -82.696441   \n",
+       "../samples/2024_how_to_focus.md                          -40.660140   \n",
+       "../samples/2024_gpt4_fake_paper.md                       -71.474606   \n",
+       "../samples/2025_h5n1_report.md                           -77.666175   \n",
+       "../samples/2024_arxiv_meh.md                             -61.776143   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...       -65.756404   \n",
+       "../samples/2024_news_anthropic.md                        -71.181132   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md          -29.038092   \n",
+       "../samples/politics_is_the_mind_killer.md                -65.298689   \n",
+       "../samples/2024_lesswrong_slop.md                        -75.973899   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...       -90.026555   \n",
+       "../samples/2024_anthropic_palintir.md                    -75.181568   \n",
+       "../samples/2025_lw_preference-inversion.md               -88.162939   \n",
+       "../samples/2024_deliberative_alignment.md                -29.432378   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...       -71.898714   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...       -85.223138   \n",
+       "../samples/2025_lw_the-intelligence-curse.md               2.610249   \n",
+       "../samples/lorem_ipsum.md                                -19.999573   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...       378.210283   \n",
+       "../samples/2024_openai_emails.md                         -95.082707   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md       144.210285   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...       -86.930632   \n",
+       "../samples/2024_bob_fanfic.md                            -35.032627   \n",
+       "../samples/2024_gwern_reddit.md                           46.703333   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...       -65.735003   \n",
+       "../samples/2024_bob_fanfic2.md                           -35.546726   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...       159.805322   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                70.526800   \n",
+       "../samples/2025_lw_review-planecrash.md                  -79.884876   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md              50.615295   \n",
+       "\n",
+       "                                                    train/before_max  \\\n",
+       "f                                                                      \n",
+       "../samples/2024_trump_appointment.md                       17.917255   \n",
+       "../samples/2024_how_to_focus.md                            13.456800   \n",
+       "../samples/2024_gpt4_fake_paper.md                         11.427372   \n",
+       "../samples/2025_h5n1_report.md                             11.928956   \n",
+       "../samples/2024_arxiv_meh.md                               13.523319   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...         12.706497   \n",
+       "../samples/2024_news_anthropic.md                          12.790111   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md            11.118690   \n",
+       "../samples/politics_is_the_mind_killer.md                  11.889731   \n",
+       "../samples/2024_lesswrong_slop.md                          15.304575   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...         12.598433   \n",
+       "../samples/2024_anthropic_palintir.md                      13.271792   \n",
+       "../samples/2025_lw_preference-inversion.md                 11.404502   \n",
+       "../samples/2024_deliberative_alignment.md                  14.758335   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...         12.878409   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...         14.798123   \n",
+       "../samples/2025_lw_the-intelligence-curse.md               17.431973   \n",
+       "../samples/lorem_ipsum.md                                  12.736030   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...         12.190395   \n",
+       "../samples/2024_openai_emails.md                           10.482502   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md         12.902117   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...         13.688265   \n",
+       "../samples/2024_bob_fanfic.md                              13.226252   \n",
+       "../samples/2024_gwern_reddit.md                            15.920402   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...         16.383120   \n",
+       "../samples/2024_bob_fanfic2.md                             13.576470   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...         13.405236   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                 11.613052   \n",
+       "../samples/2025_lw_review-planecrash.md                    14.514912   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md               13.301698   \n",
+       "\n",
+       "                                                    train/after_max  \\\n",
+       "f                                                                     \n",
+       "../samples/2024_trump_appointment.md                      11.689822   \n",
+       "../samples/2024_how_to_focus.md                           13.169463   \n",
+       "../samples/2024_gpt4_fake_paper.md                         9.812000   \n",
+       "../samples/2025_h5n1_report.md                             9.404486   \n",
+       "../samples/2024_arxiv_meh.md                              13.253697   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...        11.517975   \n",
+       "../samples/2024_news_anthropic.md                         12.667040   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md            8.321499   \n",
+       "../samples/politics_is_the_mind_killer.md                 11.973468   \n",
+       "../samples/2024_lesswrong_slop.md                         12.805290   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...        12.334391   \n",
+       "../samples/2024_anthropic_palintir.md                     12.891874   \n",
+       "../samples/2025_lw_preference-inversion.md                 8.079728   \n",
+       "../samples/2024_deliberative_alignment.md                 12.259495   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...        10.411036   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...        12.255587   \n",
+       "../samples/2025_lw_the-intelligence-curse.md              11.515382   \n",
+       "../samples/lorem_ipsum.md                                 12.230991   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...        10.575907   \n",
+       "../samples/2024_openai_emails.md                           9.602067   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md         9.158434   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...        11.766024   \n",
+       "../samples/2024_bob_fanfic.md                             11.086496   \n",
+       "../samples/2024_gwern_reddit.md                           14.091064   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...        15.471860   \n",
+       "../samples/2024_bob_fanfic2.md                            12.110069   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...        17.062244   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                11.740885   \n",
+       "../samples/2025_lw_review-planecrash.md                   10.882781   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md              12.898935   \n",
+       "\n",
+       "                                                    train/diff_max  \\\n",
+       "f                                                                    \n",
+       "../samples/2024_trump_appointment.md                     -6.227433   \n",
+       "../samples/2024_how_to_focus.md                          -0.287336   \n",
+       "../samples/2024_gpt4_fake_paper.md                       -1.615372   \n",
+       "../samples/2025_h5n1_report.md                           -2.524470   \n",
+       "../samples/2024_arxiv_meh.md                             -0.269622   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...       -1.188522   \n",
+       "../samples/2024_news_anthropic.md                        -0.123071   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md          -2.797191   \n",
+       "../samples/politics_is_the_mind_killer.md                 0.083736   \n",
+       "../samples/2024_lesswrong_slop.md                        -2.499285   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...       -0.264043   \n",
+       "../samples/2024_anthropic_palintir.md                    -0.379918   \n",
+       "../samples/2025_lw_preference-inversion.md               -3.324774   \n",
+       "../samples/2024_deliberative_alignment.md                -2.498840   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...       -2.467374   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...       -2.542537   \n",
+       "../samples/2025_lw_the-intelligence-curse.md             -5.916591   \n",
+       "../samples/lorem_ipsum.md                                -0.505038   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...       -1.614489   \n",
+       "../samples/2024_openai_emails.md                         -0.880435   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md       -3.743683   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...       -1.922241   \n",
+       "../samples/2024_bob_fanfic.md                            -2.139755   \n",
+       "../samples/2024_gwern_reddit.md                          -1.829337   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...       -0.911260   \n",
+       "../samples/2024_bob_fanfic2.md                           -1.466401   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...        3.657008   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                0.127832   \n",
+       "../samples/2025_lw_review-planecrash.md                  -3.632131   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md             -0.402762   \n",
+       "\n",
+       "                                                    train/diff%_max  \\\n",
+       "f                                                                     \n",
+       "../samples/2024_trump_appointment.md                     -34.756625   \n",
+       "../samples/2024_how_to_focus.md                           -2.135250   \n",
+       "../samples/2024_gpt4_fake_paper.md                       -14.135986   \n",
+       "../samples/2025_h5n1_report.md                           -21.162542   \n",
+       "../samples/2024_arxiv_meh.md                              -1.993755   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...        -9.353658   \n",
+       "../samples/2024_news_anthropic.md                         -0.962233   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md          -25.157557   \n",
+       "../samples/politics_is_the_mind_killer.md                  0.704275   \n",
+       "../samples/2024_lesswrong_slop.md                        -16.330311   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...        -2.095839   \n",
+       "../samples/2024_anthropic_palintir.md                     -2.862598   \n",
+       "../samples/2025_lw_preference-inversion.md               -29.153170   \n",
+       "../samples/2024_deliberative_alignment.md                -16.931722   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...       -19.158995   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...       -17.181481   \n",
+       "../samples/2025_lw_the-intelligence-curse.md             -33.941028   \n",
+       "../samples/lorem_ipsum.md                                 -3.965429   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...       -13.243940   \n",
+       "../samples/2024_openai_emails.md                          -8.399092   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md       -29.016036   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...       -14.042987   \n",
+       "../samples/2024_bob_fanfic.md                            -16.178093   \n",
+       "../samples/2024_gwern_reddit.md                          -11.490521   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...        -5.562186   \n",
+       "../samples/2024_bob_fanfic2.md                           -10.801048   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...        27.280446   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                 1.100765   \n",
+       "../samples/2025_lw_review-planecrash.md                  -25.023443   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md              -3.027902   \n",
+       "\n",
+       "                                                    train/before_sum  \\\n",
+       "f                                                                      \n",
+       "../samples/2024_trump_appointment.md                      284.586479   \n",
+       "../samples/2024_how_to_focus.md                           347.161163   \n",
+       "../samples/2024_gpt4_fake_paper.md                        349.755680   \n",
+       "../samples/2025_h5n1_report.md                            355.013296   \n",
+       "../samples/2024_arxiv_meh.md                              386.270819   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...        393.351672   \n",
+       "../samples/2024_news_anthropic.md                         424.085983   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md           424.532354   \n",
+       "../samples/politics_is_the_mind_killer.md                 446.767846   \n",
+       "../samples/2024_lesswrong_slop.md                         449.497469   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...        450.877187   \n",
+       "../samples/2024_anthropic_palintir.md                     454.109781   \n",
+       "../samples/2025_lw_preference-inversion.md                456.815627   \n",
+       "../samples/2024_deliberative_alignment.md                 467.468432   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...        471.837339   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...        477.610894   \n",
+       "../samples/2025_lw_the-intelligence-curse.md              484.856104   \n",
+       "../samples/lorem_ipsum.md                                 490.732376   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...        500.521191   \n",
+       "../samples/2024_openai_emails.md                          517.339424   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md        517.833801   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...        523.423819   \n",
+       "../samples/2024_bob_fanfic.md                             527.048090   \n",
+       "../samples/2024_gwern_reddit.md                           562.662201   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...        616.809882   \n",
+       "../samples/2024_bob_fanfic2.md                            620.444836   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...        623.402177   \n",
+       "../samples/2024_lw_the-plan-2024-update.md                673.517207   \n",
+       "../samples/2025_lw_review-planecrash.md                   684.699408   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md              801.941370   \n",
+       "\n",
+       "                                                    train/after_sum  \\\n",
+       "f                                                                     \n",
+       "../samples/2024_trump_appointment.md                     156.141833   \n",
+       "../samples/2024_how_to_focus.md                          211.773822   \n",
+       "../samples/2024_gpt4_fake_paper.md                       227.104543   \n",
+       "../samples/2025_h5n1_report.md                           198.560084   \n",
+       "../samples/2024_arxiv_meh.md                             312.432330   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...       223.485284   \n",
+       "../samples/2024_news_anthropic.md                        247.100184   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md          316.633504   \n",
+       "../samples/politics_is_the_mind_killer.md                330.564804   \n",
+       "../samples/2024_lesswrong_slop.md                        247.537046   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...       353.007281   \n",
+       "../samples/2024_anthropic_palintir.md                    310.446145   \n",
+       "../samples/2025_lw_preference-inversion.md               217.925137   \n",
+       "../samples/2024_deliberative_alignment.md                307.554861   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...       246.898066   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...       277.590406   \n",
+       "../samples/2025_lw_the-intelligence-curse.md             218.930921   \n",
+       "../samples/lorem_ipsum.md                                363.262540   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...       211.690162   \n",
+       "../samples/2024_openai_emails.md                         323.156004   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md       288.011791   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...       262.369446   \n",
+       "../samples/2024_bob_fanfic.md                            420.827773   \n",
+       "../samples/2024_gwern_reddit.md                          420.227737   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...       376.879962   \n",
+       "../samples/2024_bob_fanfic2.md                           494.885355   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...       521.238323   \n",
+       "../samples/2024_lw_the-plan-2024-update.md               536.693749   \n",
+       "../samples/2025_lw_review-planecrash.md                  345.217788   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md             504.596465   \n",
+       "\n",
+       "                                                    train/diff_sum  \\\n",
+       "f                                                                    \n",
+       "../samples/2024_trump_appointment.md                   -128.444646   \n",
+       "../samples/2024_how_to_focus.md                        -135.387340   \n",
+       "../samples/2024_gpt4_fake_paper.md                     -122.651137   \n",
+       "../samples/2025_h5n1_report.md                         -156.453212   \n",
+       "../samples/2024_arxiv_meh.md                            -73.838489   \n",
+       "../samples/2024_lw_by-default-capital-will-matt...     -169.866388   \n",
+       "../samples/2024_news_anthropic.md                      -176.985800   \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md        -107.898850   \n",
+       "../samples/politics_is_the_mind_killer.md              -116.203043   \n",
+       "../samples/2024_lesswrong_slop.md                      -201.960423   \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...      -97.869906   \n",
+       "../samples/2024_anthropic_palintir.md                  -143.663636   \n",
+       "../samples/2025_lw_preference-inversion.md             -238.890490   \n",
+       "../samples/2024_deliberative_alignment.md              -159.913571   \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...     -224.939272   \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...     -200.020488   \n",
+       "../samples/2025_lw_the-intelligence-curse.md           -265.925183   \n",
+       "../samples/lorem_ipsum.md                              -127.469837   \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...     -288.831029   \n",
+       "../samples/2024_openai_emails.md                       -194.183420   \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md     -229.822011   \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...     -261.054373   \n",
+       "../samples/2024_bob_fanfic.md                          -106.220317   \n",
+       "../samples/2024_gwern_reddit.md                        -142.434464   \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...     -239.929921   \n",
+       "../samples/2024_bob_fanfic2.md                         -125.559481   \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...     -102.163854   \n",
+       "../samples/2024_lw_the-plan-2024-update.md             -136.823458   \n",
+       "../samples/2025_lw_review-planecrash.md                -339.481621   \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md           -297.344904   \n",
+       "\n",
+       "                                                    train/diff%_sum  \n",
+       "f                                                                    \n",
+       "../samples/2024_trump_appointment.md                     -45.133784  \n",
+       "../samples/2024_how_to_focus.md                          -38.998412  \n",
+       "../samples/2024_gpt4_fake_paper.md                       -35.067661  \n",
+       "../samples/2025_h5n1_report.md                           -44.069677  \n",
+       "../samples/2024_arxiv_meh.md                             -19.115730  \n",
+       "../samples/2024_lw_by-default-capital-will-matt...       -43.184356  \n",
+       "../samples/2024_news_anthropic.md                        -41.733471  \n",
+       "../samples/2025_lw_the-laws-of-large-numbers.md          -25.415931  \n",
+       "../samples/politics_is_the_mind_killer.md                -26.009715  \n",
+       "../samples/2024_lesswrong_slop.md                        -44.930269  \n",
+       "../samples/2025_lw_parkinson-s-law-and-the-ideo...       -21.706555  \n",
+       "../samples/2024_anthropic_palintir.md                    -31.636323  \n",
+       "../samples/2025_lw_preference-inversion.md               -52.294728  \n",
+       "../samples/2024_deliberative_alignment.md                -34.208421  \n",
+       "../samples/2025_lw_the-subset-parity-learning-p...       -47.673055  \n",
+       "../samples/2025_lw_human-study-on-ai-spear-phis...       -41.879381  \n",
+       "../samples/2025_lw_the-intelligence-curse.md             -54.846207  \n",
+       "../samples/lorem_ipsum.md                                -25.975428  \n",
+       "../samples/2025_lw_the-field-of-ai-alignment-a-...       -57.706054  \n",
+       "../samples/2024_openai_emails.md                         -37.535013  \n",
+       "../samples/2025_lw_debating-buying-nvda-in-2019.md       -44.381423  \n",
+       "../samples/2025_lw_my-agi-safety-research-2024-...       -49.874378  \n",
+       "../samples/2024_bob_fanfic.md                            -20.153819  \n",
+       "../samples/2024_gwern_reddit.md                          -25.314383  \n",
+       "../samples/2025_lw_comment-on-death-and-the-gor...       -38.898521  \n",
+       "../samples/2024_bob_fanfic2.md                           -20.237010  \n",
+       "../samples/2025_lw_what-s-the-short-timeline-pl...       -16.388113  \n",
+       "../samples/2024_lw_the-plan-2024-update.md               -20.314768  \n",
+       "../samples/2025_lw_review-planecrash.md                  -49.581118  \n",
+       "../samples/2025_lw_2024-in-ai-predictions.md             -37.078135  \n",
+       "\n",
+       "[30 rows x 24 columns]"
       ]
      },
-     "execution_count": 28,
+     "execution_count": 87,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "df_res = pd.DataFrame(data)\n",
-    "df_res['len'] = df_res.content.str.len()\n",
-    "df_res = df_res[['before', 'after', 'title',  'len']].set_index('title')\n",
-    "df_res['improvement%'] = (df_res['before'] - df_res['after'])/ df_res['before']\n",
-    "df_res['improvement'] = (df_res['before'] - df_res['after'])\n",
-    "df_res['novel'] = df_res['before'] > 15\n",
-    "df_res['learnable'] = df_res['improvement%'] > 0.02\n",
-    "\n",
-    "# We can measure the final score using learnable * novel\n",
-    "# df_res['BS'] = ~df_res['learnable'] | ~df_res['novel']\n",
-    "# Or just absolute perplexity improvement\n",
-    "df_res['BS'] = df_res['improvement'] < .3\n",
-    "df_res = df_res.sort_values('improvement', ascending=False)\n",
-    "df_res"
+    "df_res.sort_values('train/before_sum').set_index('f').drop(columns=['content', 'hist', 'before', 'after', 'train/before', 'train/after', 'url'])"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 26,
+   "execution_count": 88,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1330,33 +3327,9 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 22,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "| title                                     |    before |     after |   len |   improvement% |   improvement | novel   | learnable   | BS    |\n",
-      "|:------------------------------------------|----------:|----------:|------:|---------------:|--------------:|:--------|:------------|:------|\n",
-      "| cicero from ibois, Philippe (2012-06-03). |  72.5874  |  67.7442  | 13707 |     0.066722   |     4.84318   | True    | True        | False |\n",
-      "| politics is the mind-killer               | 247.552   | 245.827   |  3158 |     0.00696722 |     1.72475   | True    | False       | False |\n",
-      "| openai board ann                          |  55.8085  |  54.6679  |  2991 |     0.0204374  |     1.14058   | True    | True        | False |\n",
-      "| How to Catch an AI Liar                   |  28.9499  |  28.0088  |  5464 |     0.0325069  |     0.941071  | True    | True        | True  |\n",
-      "| buzzfeed foi fauci emails 2023            |  23.3094  |  22.4064  | 13640 |     0.0387411  |     0.903032  | True    | True        | True  |\n",
-      "| Gemini to Q*                              |  11.7564  |  11.1906  | 42604 |     0.0481219  |     0.56574   | False   | True        | True  |\n",
-      "| LK-99-en                                  |  14.5138  |  14.0661  | 15432 |     0.03085    |     0.447752  | False   | True        | True  |\n",
-      "| LK-99-es                                  |  11.415   |  10.9729  | 12970 |     0.0387271  |     0.44207   | False   | True        | True  |\n",
-      "| disney appointment                        | 118.826   | 118.42    |  3653 |     0.003417   |     0.406029  | True    | False       | True  |\n",
-      "| weak to strong                            |  46.7642  |  46.4047  |  5811 |     0.00768638 |     0.359447  | True    | False       | True  |\n",
-      "| blechley declaration                      |  17.8691  |  17.5242  |  7762 |     0.0193007  |     0.344887  | True    | False       | True  |\n",
-      "| Lorem ipsum                               |   6.56484 |   6.26879 | 19649 |     0.0450961  |     0.296049  | False   | True        | True  |\n",
-      "| statement by whitehouse on passing        |  29.1971  |  28.9397  |  1641 |     0.00881732 |     0.257441  | True    | False       | True  |\n",
-      "| harvard announcment caplain israel hamas  |  45.3474  |  45.1273  |  4247 |     0.00485323 |     0.220081  | True    | False       | True  |\n",
-      "| fake ai hoax paper                        |   7.76698 |   7.69723 |  3290 |     0.00898037 |     0.0697503 | False   | False       | True  |\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
     "print(df_res.to_markdown())"
    ]
@@ -1425,52 +3398,9 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 25,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/generation/utils.py:1421: UserWarning: You have modified the pretrained model configuration to control generation. This is a deprecated strategy to control generation and will be removed soon, in a future version. Please use and modify the model generation configuration (see https://huggingface.co/docs/transformers/generation_strategies#default-text-generation-configuration )\n",
-      "  warnings.warn(\n"
-     ]
-    },
-    {
-     "ename": "OutOfMemoryError",
-     "evalue": "CUDA out of memory. Tried to allocate 1.07 GiB. GPU 0 has a total capacty of 10.75 GiB of which 612.62 MiB is free. Including non-PyTorch memory, this process has 10.15 GiB memory in use. Of the allocated memory 9.78 GiB is allocated by PyTorch, and 179.66 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF",
-     "output_type": "error",
-     "traceback": [
-      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
-      "\u001b[0;31mOutOfMemoryError\u001b[0m                          Traceback (most recent call last)",
-      "Cell \u001b[0;32mIn[25], line 2\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[38;5;28;01mwith\u001b[39;00m model\u001b[38;5;241m.\u001b[39mdisable_adapter():\n\u001b[0;32m----> 2\u001b[0m     \u001b[43mgen\u001b[49m\u001b[43m(\u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mds_train\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mwith_format\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[38;5;124;43mpt\u001b[39;49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[43m)\u001b[49m\u001b[43m[\u001b[49m\u001b[38;5;241;43m0\u001b[39;49m\u001b[43m]\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtokenizer\u001b[49m\u001b[43m)\u001b[49m\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/utils/_contextlib.py:115\u001b[0m, in \u001b[0;36mcontext_decorator.<locals>.decorate_context\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    112\u001b[0m \u001b[38;5;129m@functools\u001b[39m\u001b[38;5;241m.\u001b[39mwraps(func)\n\u001b[1;32m    113\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mdecorate_context\u001b[39m(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[1;32m    114\u001b[0m     \u001b[38;5;28;01mwith\u001b[39;00m ctx_factory():\n\u001b[0;32m--> 115\u001b[0m         \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mfunc\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
-      "Cell \u001b[0;32mIn[23], line 6\u001b[0m, in \u001b[0;36mgen\u001b[0;34m(model, inputs, tokenizer, clean)\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[38;5;129m@torch\u001b[39m\u001b[38;5;241m.\u001b[39mno_grad()\n\u001b[1;32m      5\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mgen\u001b[39m(model, inputs, tokenizer, clean\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m):\n\u001b[0;32m----> 6\u001b[0m     s \u001b[38;5;241m=\u001b[39m \u001b[43mmodel\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mgenerate\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m      7\u001b[0m \u001b[43m        \u001b[49m\u001b[43minput_ids\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43minputs\u001b[49m\u001b[43m[\u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43minput_ids\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m]\u001b[49m\u001b[43m[\u001b[49m\u001b[38;5;28;43;01mNone\u001b[39;49;00m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43m:\u001b[49m\u001b[43m]\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mto\u001b[49m\u001b[43m(\u001b[49m\u001b[43mmodel\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mdevice\u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m      8\u001b[0m \u001b[43m        \u001b[49m\u001b[43mattention_mask\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43minputs\u001b[49m\u001b[43m[\u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mattention_mask\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m]\u001b[49m\u001b[43m[\u001b[49m\u001b[38;5;28;43;01mNone\u001b[39;49;00m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43m:\u001b[49m\u001b[43m]\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mto\u001b[49m\u001b[43m(\u001b[49m\u001b[43mmodel\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mdevice\u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m      9\u001b[0m \u001b[43m        \u001b[49m\u001b[43muse_cache\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m     10\u001b[0m \u001b[43m        \u001b[49m\u001b[43mmax_new_tokens\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;241;43m100\u001b[39;49m\u001b[43m,\u001b[49m\n\u001b[1;32m     11\u001b[0m \u001b[43m        \u001b[49m\u001b[43mmin_new_tokens\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;241;43m100\u001b[39;49m\u001b[43m,\u001b[49m\n\u001b[1;32m     12\u001b[0m \u001b[43m        \u001b[49m\u001b[43mdo_sample\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m     13\u001b[0m \u001b[43m        \u001b[49m\u001b[43mearly_stopping\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m     14\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m     15\u001b[0m     input_l \u001b[38;5;241m=\u001b[39m inputs[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124minput_ids\u001b[39m\u001b[38;5;124m\"\u001b[39m]\u001b[38;5;241m.\u001b[39mshape[\u001b[38;5;241m0\u001b[39m]\n\u001b[1;32m     16\u001b[0m     tokenizer_kwargs\u001b[38;5;241m=\u001b[39m\u001b[38;5;28mdict\u001b[39m(clean_up_tokenization_spaces\u001b[38;5;241m=\u001b[39mclean, skip_special_tokens\u001b[38;5;241m=\u001b[39mclean)\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/utils/_contextlib.py:115\u001b[0m, in \u001b[0;36mcontext_decorator.<locals>.decorate_context\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    112\u001b[0m \u001b[38;5;129m@functools\u001b[39m\u001b[38;5;241m.\u001b[39mwraps(func)\n\u001b[1;32m    113\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mdecorate_context\u001b[39m(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[1;32m    114\u001b[0m     \u001b[38;5;28;01mwith\u001b[39;00m ctx_factory():\n\u001b[0;32m--> 115\u001b[0m         \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mfunc\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/generation/utils.py:1606\u001b[0m, in \u001b[0;36mGenerationMixin.generate\u001b[0;34m(self, inputs, generation_config, logits_processor, stopping_criteria, prefix_allowed_tokens_fn, synced_gpus, assistant_model, streamer, negative_prompt_ids, negative_prompt_attention_mask, **kwargs)\u001b[0m\n\u001b[1;32m   1589\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39massisted_decoding(\n\u001b[1;32m   1590\u001b[0m         input_ids,\n\u001b[1;32m   1591\u001b[0m         assistant_model\u001b[38;5;241m=\u001b[39massistant_model,\n\u001b[0;32m   (...)\u001b[0m\n\u001b[1;32m   1602\u001b[0m         \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mmodel_kwargs,\n\u001b[1;32m   1603\u001b[0m     )\n\u001b[1;32m   1604\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m generation_mode \u001b[38;5;241m==\u001b[39m GenerationMode\u001b[38;5;241m.\u001b[39mGREEDY_SEARCH:\n\u001b[1;32m   1605\u001b[0m     \u001b[38;5;66;03m# 11. run greedy search\u001b[39;00m\n\u001b[0;32m-> 1606\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mgreedy_search\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m   1607\u001b[0m \u001b[43m        \u001b[49m\u001b[43minput_ids\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1608\u001b[0m \u001b[43m        \u001b[49m\u001b[43mlogits_processor\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mlogits_processor\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1609\u001b[0m \u001b[43m        \u001b[49m\u001b[43mstopping_criteria\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mstopping_criteria\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1610\u001b[0m \u001b[43m        \u001b[49m\u001b[43mpad_token_id\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mgeneration_config\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mpad_token_id\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1611\u001b[0m \u001b[43m        \u001b[49m\u001b[43meos_token_id\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mgeneration_config\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43meos_token_id\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1612\u001b[0m \u001b[43m        \u001b[49m\u001b[43moutput_scores\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mgeneration_config\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43moutput_scores\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1613\u001b[0m \u001b[43m        \u001b[49m\u001b[43mreturn_dict_in_generate\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mgeneration_config\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mreturn_dict_in_generate\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1614\u001b[0m \u001b[43m        \u001b[49m\u001b[43msynced_gpus\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43msynced_gpus\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1615\u001b[0m \u001b[43m        \u001b[49m\u001b[43mstreamer\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mstreamer\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1616\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mmodel_kwargs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1617\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1619\u001b[0m \u001b[38;5;28;01melif\u001b[39;00m generation_mode \u001b[38;5;241m==\u001b[39m GenerationMode\u001b[38;5;241m.\u001b[39mCONTRASTIVE_SEARCH:\n\u001b[1;32m   1620\u001b[0m     \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m model_kwargs[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124muse_cache\u001b[39m\u001b[38;5;124m\"\u001b[39m]:\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/transformers/generation/utils.py:2454\u001b[0m, in \u001b[0;36mGenerationMixin.greedy_search\u001b[0;34m(self, input_ids, logits_processor, stopping_criteria, max_length, pad_token_id, eos_token_id, output_attentions, output_hidden_states, output_scores, return_dict_in_generate, synced_gpus, streamer, **model_kwargs)\u001b[0m\n\u001b[1;32m   2451\u001b[0m model_inputs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mprepare_inputs_for_generation(input_ids, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mmodel_kwargs)\n\u001b[1;32m   2453\u001b[0m \u001b[38;5;66;03m# forward pass to get next token\u001b[39;00m\n\u001b[0;32m-> 2454\u001b[0m outputs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m   2455\u001b[0m \u001b[43m    \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mmodel_inputs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   2456\u001b[0m \u001b[43m    \u001b[49m\u001b[43mreturn_dict\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m   2457\u001b[0m \u001b[43m    \u001b[49m\u001b[43moutput_attentions\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43moutput_attentions\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   2458\u001b[0m \u001b[43m    \u001b[49m\u001b[43moutput_hidden_states\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43moutput_hidden_states\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   2459\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   2461\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m synced_gpus \u001b[38;5;129;01mand\u001b[39;00m this_peer_finished:\n\u001b[1;32m   2462\u001b[0m     \u001b[38;5;28;01mcontinue\u001b[39;00m  \u001b[38;5;66;03m# don't waste resources running the code we don't need\u001b[39;00m\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/nn/modules/module.py:1518\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1516\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)  \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m   1517\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1518\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/nn/modules/module.py:1527\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1522\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1523\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m   1524\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m   1525\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m   1526\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1527\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1529\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[1;32m   1530\u001b[0m     result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n",
-      "File \u001b[0;32m~/.cache/huggingface/modules/transformers_modules/microsoft/phi-2/d3186761bf5c4409f7679359284066c25ab668ee/modeling_phi.py:953\u001b[0m, in \u001b[0;36mPhiForCausalLM.forward\u001b[0;34m(self, input_ids, past_key_values, attention_mask, labels, **kwargs)\u001b[0m\n\u001b[1;32m    945\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mforward\u001b[39m(\n\u001b[1;32m    946\u001b[0m     \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m    947\u001b[0m     input_ids: torch\u001b[38;5;241m.\u001b[39mLongTensor,\n\u001b[0;32m   (...)\u001b[0m\n\u001b[1;32m    951\u001b[0m     \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs,\n\u001b[1;32m    952\u001b[0m ) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m CausalLMOutputWithPast:\n\u001b[0;32m--> 953\u001b[0m     hidden_states \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mtransformer\u001b[49m\u001b[43m(\u001b[49m\u001b[43minput_ids\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mpast_key_values\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mpast_key_values\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mattention_mask\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mattention_mask\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    954\u001b[0m     lm_logits \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mlm_head(hidden_states)\n\u001b[1;32m    956\u001b[0m     loss \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/nn/modules/module.py:1518\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1516\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)  \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m   1517\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1518\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/nn/modules/module.py:1527\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1522\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1523\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m   1524\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m   1525\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m   1526\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1527\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1529\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[1;32m   1530\u001b[0m     result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n",
-      "File \u001b[0;32m~/.cache/huggingface/modules/transformers_modules/microsoft/phi-2/d3186761bf5c4409f7679359284066c25ab668ee/modeling_phi.py:915\u001b[0m, in \u001b[0;36mPhiModel.forward\u001b[0;34m(self, input_ids, past_key_values, attention_mask)\u001b[0m\n\u001b[1;32m    912\u001b[0m hidden_states \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39membd(input_ids)\n\u001b[1;32m    914\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m layer \u001b[38;5;129;01min\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mh:\n\u001b[0;32m--> 915\u001b[0m     hidden_states \u001b[38;5;241m=\u001b[39m \u001b[43mlayer\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m    916\u001b[0m \u001b[43m        \u001b[49m\u001b[43mhidden_states\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    917\u001b[0m \u001b[43m        \u001b[49m\u001b[43mpast_key_values\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mpast_key_values\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    918\u001b[0m \u001b[43m        \u001b[49m\u001b[43mattention_mask\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mattention_mask\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    919\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    921\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m hidden_states\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/nn/modules/module.py:1518\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1516\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)  \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m   1517\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1518\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/nn/modules/module.py:1527\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1522\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1523\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m   1524\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m   1525\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m   1526\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1527\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1529\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[1;32m   1530\u001b[0m     result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n",
-      "File \u001b[0;32m~/.cache/huggingface/modules/transformers_modules/microsoft/phi-2/d3186761bf5c4409f7679359284066c25ab668ee/modeling_phi.py:770\u001b[0m, in \u001b[0;36mParallelBlock.forward\u001b[0;34m(self, hidden_states, past_key_values, attention_mask, **kwargs)\u001b[0m\n\u001b[1;32m    767\u001b[0m residual \u001b[38;5;241m=\u001b[39m hidden_states\n\u001b[1;32m    768\u001b[0m hidden_states \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mln(hidden_states)\n\u001b[0;32m--> 770\u001b[0m attn_outputs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mmixer\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m    771\u001b[0m \u001b[43m    \u001b[49m\u001b[43mhidden_states\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    772\u001b[0m \u001b[43m    \u001b[49m\u001b[43mpast_key_values\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mpast_key_values\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    773\u001b[0m \u001b[43m    \u001b[49m\u001b[43mattention_mask\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mattention_mask\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    774\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    775\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(attn_outputs, \u001b[38;5;28mtuple\u001b[39m):\n\u001b[1;32m    776\u001b[0m     attn_outputs \u001b[38;5;241m=\u001b[39m attn_outputs[\u001b[38;5;241m0\u001b[39m]\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/nn/modules/module.py:1518\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1516\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)  \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m   1517\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1518\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/nn/modules/module.py:1527\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1522\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1523\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m   1524\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m   1525\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m   1526\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1527\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1529\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[1;32m   1530\u001b[0m     result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n",
-      "File \u001b[0;32m~/.cache/huggingface/modules/transformers_modules/microsoft/phi-2/d3186761bf5c4409f7679359284066c25ab668ee/modeling_phi.py:726\u001b[0m, in \u001b[0;36mMHA.forward\u001b[0;34m(self, x, past_key_values, attention_mask, **kwargs)\u001b[0m\n\u001b[1;32m    722\u001b[0m         attn_output \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_self_attn(x, attention_mask)\n\u001b[1;32m    723\u001b[0m     \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m    724\u001b[0m         \u001b[38;5;66;03m# If `past_key_values` are supplied, it means that we might have cached values and\u001b[39;00m\n\u001b[1;32m    725\u001b[0m         \u001b[38;5;66;03m# could take advantage of cross-attention\u001b[39;00m\n\u001b[0;32m--> 726\u001b[0m         attn_output \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_forward_cross_attn\u001b[49m\u001b[43m(\u001b[49m\u001b[43mx\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mpast_key_values\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mattention_mask\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    727\u001b[0m \u001b[38;5;66;03m# MQA / GQA\u001b[39;00m\n\u001b[1;32m    728\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m    729\u001b[0m     \u001b[38;5;66;03m# Regardless of `past_key_values` being supplied or not, it always use cross-attention\u001b[39;00m\n\u001b[1;32m    730\u001b[0m     \u001b[38;5;66;03m# because `q` and `kv` lengths might be different\u001b[39;00m\n\u001b[1;32m    731\u001b[0m     attn_output \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_cross_attn(x, past_key_values, attention_mask)\n",
-      "File \u001b[0;32m~/.cache/huggingface/modules/transformers_modules/microsoft/phi-2/d3186761bf5c4409f7679359284066c25ab668ee/modeling_phi.py:704\u001b[0m, in \u001b[0;36mMHA._forward_cross_attn\u001b[0;34m(self, x, past_key_values, key_padding_mask)\u001b[0m\n\u001b[1;32m    695\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mcheckpointing:\n\u001b[1;32m    696\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m torch\u001b[38;5;241m.\u001b[39mutils\u001b[38;5;241m.\u001b[39mcheckpoint\u001b[38;5;241m.\u001b[39mcheckpoint(\n\u001b[1;32m    697\u001b[0m         \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39minner_cross_attn,\n\u001b[1;32m    698\u001b[0m         q,\n\u001b[0;32m   (...)\u001b[0m\n\u001b[1;32m    701\u001b[0m         causal\u001b[38;5;241m=\u001b[39mcausal,\n\u001b[1;32m    702\u001b[0m     )\n\u001b[0;32m--> 704\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43minner_cross_attn\u001b[49m\u001b[43m(\u001b[49m\u001b[43mq\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mkv\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mkey_padding_mask\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mkey_padding_mask\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mcausal\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mcausal\u001b[49m\u001b[43m)\u001b[49m\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/nn/modules/module.py:1518\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1516\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)  \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m   1517\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1518\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/nn/modules/module.py:1527\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1522\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1523\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m   1524\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m   1525\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m   1526\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1527\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1529\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[1;32m   1530\u001b[0m     result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/amp/autocast_mode.py:16\u001b[0m, in \u001b[0;36mautocast_decorator.<locals>.decorate_autocast\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     13\u001b[0m \u001b[38;5;129m@functools\u001b[39m\u001b[38;5;241m.\u001b[39mwraps(func)\n\u001b[1;32m     14\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mdecorate_autocast\u001b[39m(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[1;32m     15\u001b[0m     \u001b[38;5;28;01mwith\u001b[39;00m autocast_instance:\n\u001b[0;32m---> 16\u001b[0m         \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mfunc\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
-      "File \u001b[0;32m/media/wassname/SGIronWolf/projects5/bs_writing_detector/.venv/lib/python3.11/site-packages/torch/amp/autocast_mode.py:16\u001b[0m, in \u001b[0;36mautocast_decorator.<locals>.decorate_autocast\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     13\u001b[0m \u001b[38;5;129m@functools\u001b[39m\u001b[38;5;241m.\u001b[39mwraps(func)\n\u001b[1;32m     14\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mdecorate_autocast\u001b[39m(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[1;32m     15\u001b[0m     \u001b[38;5;28;01mwith\u001b[39;00m autocast_instance:\n\u001b[0;32m---> 16\u001b[0m         \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mfunc\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
-      "File \u001b[0;32m~/.cache/huggingface/modules/transformers_modules/microsoft/phi-2/d3186761bf5c4409f7679359284066c25ab668ee/modeling_phi.py:452\u001b[0m, in \u001b[0;36mCrossAttention.forward\u001b[0;34m(self, q, kv, causal, key_padding_mask, **kwargs)\u001b[0m\n\u001b[1;32m    448\u001b[0m     causal_mask \u001b[38;5;241m=\u001b[39m cols \u001b[38;5;241m>\u001b[39m rows \u001b[38;5;241m+\u001b[39m seqlen_k \u001b[38;5;241m-\u001b[39m seqlen_q\n\u001b[1;32m    450\u001b[0m     scores \u001b[38;5;241m=\u001b[39m scores\u001b[38;5;241m.\u001b[39mmasked_fill(causal_mask, \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m10000.0\u001b[39m)\n\u001b[0;32m--> 452\u001b[0m attention \u001b[38;5;241m=\u001b[39m \u001b[43mtorch\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43msoftmax\u001b[49m\u001b[43m(\u001b[49m\u001b[43mscores\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mdim\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;241;43m-\u001b[39;49m\u001b[38;5;241;43m1\u001b[39;49m\u001b[43m)\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mto\u001b[49m\u001b[43m(\u001b[49m\u001b[43mv\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mdtype\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    453\u001b[0m attention \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mdrop(attention)\n\u001b[1;32m    455\u001b[0m output \u001b[38;5;241m=\u001b[39m torch\u001b[38;5;241m.\u001b[39meinsum(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mbhts,bshd->bthd\u001b[39m\u001b[38;5;124m\"\u001b[39m, attention, v)\n",
-      "\u001b[0;31mOutOfMemoryError\u001b[0m: CUDA out of memory. Tried to allocate 1.07 GiB. GPU 0 has a total capacty of 10.75 GiB of which 612.62 MiB is free. Including non-PyTorch memory, this process has 10.15 GiB memory in use. Of the allocated memory 9.78 GiB is allocated by PyTorch, and 179.66 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
     "with model.disable_adapter():\n",
     "    gen(model, ds_train.with_format('pt')[0], tokenizer)"
diff --git a/nbs/02_detection_using_tldr_prompt.ipynb b/nbs/02_detection_using_tldr_prompt.ipynb
index fe7d525..1f29d1a 100644
--- a/nbs/02_detection_using_tldr_prompt.ipynb
+++ b/nbs/02_detection_using_tldr_prompt.ipynb
@@ -55,139 +55,51 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 30,
    "metadata": {},
    "outputs": [
     {
-     "name": "stdout",
+     "name": "stderr",
      "output_type": "stream",
      "text": [
-      "../samples/2024_anthropic_palintir.md\n",
-      "../samples/2024_arxiv_meh.md\n",
-      "../samples/2024_bob_fanfic.md\n",
-      "../samples/2024_bob_fanfic2.md\n",
-      "../samples/2024_deliberative_alignment.md\n",
-      "../samples/2024_gpt4_fake_paper.md\n",
-      "../samples/2024_gwern_reddit.md\n",
-      "../samples/2024_how_to_focus.md\n",
-      "../samples/2024_lesswrong_slop.md\n",
-      "../samples/2024_news_anthropic.md\n",
-      "../samples/2024_openai_emails.md\n",
-      "../samples/2024_trump_appointment.md\n",
-      "../samples/2025_h5n1_report.md\n",
-      "../samples/2025_lw_human-study-on-ai-spear-phishing-campaigns.md\n",
-      "../samples/2025_lw_parkinson-s-law-and-the-ideology-of-statistics-1.md\n",
-      "../samples/2025_lw_the-intelligence-curse.md\n",
-      "../samples/2025_lw_the-laws-of-large-numbers.md\n",
-      "../samples/2025_lw_what-s-the-short-timeline-plan.md\n",
-      "../samples/lorem_ipsum.md\n",
-      "../samples/politics_is_the_mind_killer.md\n"
+      "\u001b[32m2025-01-06 11:35:53.921\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_anthropic_palintir.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.921\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_arxiv_meh.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.922\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_bob_fanfic.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.922\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_bob_fanfic2.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.922\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_deliberative_alignment.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.940\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_gpt4_fake_paper.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.941\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_gwern_reddit.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.941\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_how_to_focus.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.942\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_lesswrong_slop.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.942\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_lw_by-default-capital-will-matter-more-than-ever-after-agi.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.959\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_lw_the-plan-2024-update.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.960\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_news_anthropic.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.961\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_openai_emails.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.973\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2024_trump_appointment.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.974\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_h5n1_report.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.974\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_2024-in-ai-predictions.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.992\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_comment-on-death-and-the-gorgon.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.992\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_debating-buying-nvda-in-2019.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:53.993\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_human-study-on-ai-spear-phishing-campaigns.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:54.000\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_my-agi-safety-research-2024-review-25-plans.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:54.002\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_parkinson-s-law-and-the-ideology-of-statistics-1.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:54.002\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_preference-inversion.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:54.003\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_review-planecrash.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:54.003\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_the-field-of-ai-alignment-a-postmortem-and-what-to-do-about.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:54.004\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_the-intelligence-curse.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:54.004\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_the-laws-of-large-numbers.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:54.004\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_the-subset-parity-learning-problem-much-more-than-you-wanted.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:54.005\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/2025_lw_what-s-the-short-timeline-plan.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:54.006\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/lorem_ipsum.md\u001b[0m\n",
+      "\u001b[32m2025-01-06 11:35:54.006\u001b[0m | \u001b[34m\u001b[1mDEBUG   \u001b[0m | \u001b[36mbs_writing_detector.data.load_md\u001b[0m:\u001b[36mload_md\u001b[0m:\u001b[36m10\u001b[0m - \u001b[34m\u001b[1m../samples/politics_is_the_mind_killer.md\u001b[0m\n"
      ]
-    },
-    {
-     "data": {
-      "text/html": [
-       "<div>\n",
-       "<style scoped>\n",
-       "    .dataframe tbody tr th:only-of-type {\n",
-       "        vertical-align: middle;\n",
-       "    }\n",
-       "\n",
-       "    .dataframe tbody tr th {\n",
-       "        vertical-align: top;\n",
-       "    }\n",
-       "\n",
-       "    .dataframe thead th {\n",
-       "        text-align: right;\n",
-       "    }\n",
-       "</style>\n",
-       "<table border=\"1\" class=\"dataframe\">\n",
-       "  <thead>\n",
-       "    <tr style=\"text-align: right;\">\n",
-       "      <th></th>\n",
-       "      <th>title</th>\n",
-       "      <th>f</th>\n",
-       "      <th>content</th>\n",
-       "      <th>url</th>\n",
-       "      <th>novelty</th>\n",
-       "      <th>date</th>\n",
-       "      <th>in_training</th>\n",
-       "    </tr>\n",
-       "  </thead>\n",
-       "  <tbody>\n",
-       "    <tr>\n",
-       "      <th>0</th>\n",
-       "      <td>Anthropic and Palantir Partner to Bring Claude...</td>\n",
-       "      <td>../samples/2024_anthropic_palintir.md</td>\n",
-       "      <td>Anthropic and Palantir Technologies Inc. (NYSE...</td>\n",
-       "      <td>https://investors.palantir.com/news-details/20...</td>\n",
-       "      <td>0.20</td>\n",
-       "      <td>2024-07-11 00:00:00+00:00</td>\n",
-       "      <td>False</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>1</th>\n",
-       "      <td>TradingAgents: Multi-Agents LLM Financial Trad...</td>\n",
-       "      <td>../samples/2024_arxiv_meh.md</td>\n",
-       "      <td>TradingAgents: Multi-Agents LLM Financial Trad...</td>\n",
-       "      <td>https://arxiv.org/html/2412.20138v1</td>\n",
-       "      <td>0.15</td>\n",
-       "      <td>2024-12-28 00:00:00+00:00</td>\n",
-       "      <td>False</td>\n",
-       "    </tr>\n",
-       "  </tbody>\n",
-       "</table>\n",
-       "</div>"
-      ],
-      "text/plain": [
-       "                                               title  \\\n",
-       "0  Anthropic and Palantir Partner to Bring Claude...   \n",
-       "1  TradingAgents: Multi-Agents LLM Financial Trad...   \n",
-       "\n",
-       "                                       f  \\\n",
-       "0  ../samples/2024_anthropic_palintir.md   \n",
-       "1           ../samples/2024_arxiv_meh.md   \n",
-       "\n",
-       "                                             content  \\\n",
-       "0  Anthropic and Palantir Technologies Inc. (NYSE...   \n",
-       "1  TradingAgents: Multi-Agents LLM Financial Trad...   \n",
-       "\n",
-       "                                                 url  novelty  \\\n",
-       "0  https://investors.palantir.com/news-details/20...     0.20   \n",
-       "1                https://arxiv.org/html/2412.20138v1     0.15   \n",
-       "\n",
-       "                       date  in_training  \n",
-       "0 2024-07-11 00:00:00+00:00        False  \n",
-       "1 2024-12-28 00:00:00+00:00        False  "
-      ]
-     },
-     "execution_count": 4,
-     "metadata": {},
-     "output_type": "execute_result"
     }
    ],
    "source": [
-    "import frontmatter\n",
+    "from bs_writing_detector.data.load_md import load_md_df\n",
     "from pathlib import Path\n",
-    "from loguru import logger\n",
     "\n",
-    "sample_files = sorted(Path(\"../samples/\").glob('*.md'))\n",
-    "# print(sample_files)\n",
-    "\n",
-    "max_chars = 2000\n",
-    "max_summary_frac = 0.1\n",
-    "samples = []\n",
-    "for f in sample_files:\n",
-    "    print(f)\n",
-    "    with open(f, \"r\") as file:\n",
-    "        post = frontmatter.load(file)\n",
-    "        samples.append(dict(content=post.content[:max_chars], f=f, **post.metadata))\n",
-    "\n",
-    "df = pd.DataFrame(samples)\n",
-    "df = df[['title', 'f', 'content', 'url', 'novelty', 'date']]\n",
-    "df['date'] = pd.to_datetime(df['date'], utc=True)\n",
-    "df['in_training'] = df.date < '2024-01-01'\n",
-    "df.head(2)"
+    "df = load_md_df(Path(\"../samples/\"))"
    ]
   },
   {
@@ -195,7 +107,10 @@
    "execution_count": null,
    "metadata": {},
    "outputs": [],
-   "source": []
+   "source": [
+    "max_summary_frac    = 0.1\n",
+    "max_chars = 2000"
+   ]
   },
   {
    "cell_type": "markdown",
@@ -206,7 +121,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 9,
    "metadata": {},
    "outputs": [
     {
@@ -215,7 +130,7 @@
        "True"
       ]
      },
-     "execution_count": 5,
+     "execution_count": 9,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -228,15 +143,15 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
+   "execution_count": 10,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "TradingAgents framework uses multi-agent LLM for financial trading, mimicking real trading dynamics. Roles include analysts and traders with diverse risk profiles. Key features include market conditio\n",
-      "n assessors, risk management, and decision-making. Superior to baseline models, improves returns, Sharpe ratio, max drawdown.\n"
+      "Academic paper titled \"TradingAgents: Multi-Agents LLM Financial Trading Framework\". Authors propose new framework using large language model (LLM) powered agents for collaborative finance trading. Mu\n",
+      "lti-agent system mirrors trading firm dynamics with analysts & traders roles, improving performance. Highlights advantages over traditional models.\n"
      ]
     },
     {
@@ -245,7 +160,7 @@
        "(None, None)"
       ]
      },
-     "execution_count": 6,
+     "execution_count": 10,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -303,14 +218,14 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 11,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "TradingAgents introduces a multi-agent financial trading framework using large language models. Agents specialize in roles like risk analysis and market condition evaluation, improving trading performance and showing potential for LLM frameworks in finance.\n"
+      "TradingAgents is a novel multi-agent framework for financial trading powered by large language models (LLMs), aiming to replicate the dynamics of real-world trading firms to enhance trading performance.\n"
      ]
     }
    ],
@@ -321,16 +236,17 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 8,
+   "execution_count": 12,
    "metadata": {},
    "outputs": [],
    "source": [
+    "from bs_writing_detector.data.load_md import load_md_df\n",
     "from bs_writing_detector.metrics.ppx import perplexity_compute"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 9,
+   "execution_count": 13,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -353,7 +269,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 10,
+   "execution_count": 14,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -373,7 +289,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 11,
+   "execution_count": 15,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -391,13 +307,10 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 12,
+   "execution_count": 16,
    "metadata": {},
    "outputs": [],
-   "source": [
-    "\n",
-    "from bs_writing_detector.metrics.ppx import perplexity_compute"
-   ]
+   "source": []
   },
   {
    "cell_type": "code",
@@ -408,7 +321,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 17,
    "metadata": {},
    "outputs": [
     {
@@ -429,12 +342,12 @@
     {
      "data": {
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "1d28b339f7c3495c97adef2c4f1938ad",
+       "model_id": "8d4a2d678a0c45f39061c7c07defe0fa",
        "version_major": 2,
        "version_minor": 0
       },
       "text/plain": [
-       "  0%|          | 0/20 [00:00<?, ?it/s]"
+       "  0%|          | 0/30 [00:00<?, ?it/s]"
       ]
      },
      "metadata": {},
@@ -444,8 +357,36 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "2.731894616981088 2.464229501783848 | Anthropic and Palantir Partner to Bring Claude AI Models to AWS for U.S. Government Intelligence and Defense Operations\n",
-      "3.018628165803172 2.8031317269160456 | TradingAgents: Multi-Agents LLM Financial Trading Framework\n"
+      "2.731894616981088 2.4346583814670644 | Anthropic and Palantir Partner to Bring Claude AI Models to AWS for U.S. Government Intelligence and Defense Operations\n",
+      "3.018628165803172 2.787076804860793 | TradingAgents: Multi-Agents LLM Financial Trading Framework\n",
+      "3.1768678330086373 3.0573941246602545 | Flower Crowns and Furry Mishaps by MyPalAI\n",
+      "3.7198927586704955 3.619981446455229 | Paradox's Box (Bobiverse) by Mark4man\n",
+      "3.22479815838576 3.0429821194746554 | Deliberative Alignment: Reasoning Enables Safer Language Models\n",
+      "2.341998207471676 2.09474255727685 | fake ai hoax paper made up by gpt-4\n",
+      "3.66733124948317 3.466211730036242 | Hardware Hedging Against Scaling Regime Shifts (self.mlscaling)\n",
+      "1.7733189346720872 1.6321287018846171 | How to Focus\n",
+      "2.7362908764318985 2.4942711856630115 | Deontic Explorations In \"Paying To Talk To Slaves\"\n",
+      "2.9917198510817538 2.809688214010793 | By default, capital will matter more than ever after AGI\n",
+      "2.8518650030297987 2.7567292506743093 | The Plan - 2024 Update\n",
+      "2.5235848208836145 2.3560869528313555 | Amazon-backed Anthropic debuts AI agents that can do complex tasks, racing against OpenAI, Microsoft and Google\n",
+      "2.92210851534712 2.78284017383792 | OpenAI Email Archives from Musk v. Altman\n",
+      "2.8115177075068156 2.595940358411227 | President Trump Announces Morgan Ortagus as Deputy Special Presidential Envoy for Middle East Peace\n",
+      "2.616611967668609 2.4271711949948913 | CDC Report on Missouri H5N1 Serology Testing\n",
+      "3.6667002914509097 3.5698842933939248 | 2024 in AI predictions\n",
+      "3.3718713704954113 3.177144504219489 | Comment on 'Death and the Gorgon'\n",
+      "2.9332370335745703 2.7845472265409943 | debating buying NVDA in 2019\n",
+      "2.849587789907025 2.5270508773518965 | Human study on AI spear phishing campaigns\n",
+      "2.705293411980132 2.5398094839308842 | My AGI safety research—2024 review, ’25 plans\n",
+      "3.2153463730924123 2.927131202382346 | Parkinson's Law and the Ideology of Statistics\n",
+      "3.3662234347041062 3.222519686042247 | Preference Inversion\n",
+      "3.105758021576236 2.8918632517341805 | Review: Planecrash\n",
+      "2.901739284029148 2.7977401416483323 | The Field of AI Alignment: A Postmortem, and What To Do About It\n",
+      "3.326114864740971 3.2126946656100723 | The Intelligence Curse\n",
+      "2.630559911577212 2.5053266023730374 | The Laws of Large Numbers\n",
+      "2.6887638201223356 2.549592624455343 | The subset parity learning problem: much more than you wanted to know\n",
+      "3.330626070877462 3.03754643548893 | What’s the short timeline plan?\n",
+      "2.0824154792229335 2.07074254503266 | Lorem ipsum\n",
+      "2.870609588993406 2.8476737625085122 | politics is the mind-killer\n"
      ]
     }
    ],
@@ -508,9 +449,1160 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 18,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>before_mean</th>\n",
+       "      <th>before_sum</th>\n",
+       "      <th>before_std</th>\n",
+       "      <th>before_min</th>\n",
+       "      <th>before_max</th>\n",
+       "      <th>before_ppx</th>\n",
+       "      <th>after_mean</th>\n",
+       "      <th>after_sum</th>\n",
+       "      <th>after_std</th>\n",
+       "      <th>after_min</th>\n",
+       "      <th>...</th>\n",
+       "      <th>model</th>\n",
+       "      <th>summary</th>\n",
+       "      <th>len</th>\n",
+       "      <th>n_tokens</th>\n",
+       "      <th>f</th>\n",
+       "      <th>content</th>\n",
+       "      <th>url</th>\n",
+       "      <th>novelty</th>\n",
+       "      <th>date</th>\n",
+       "      <th>in_training</th>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>title</th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>Anthropic and Palantir Partner to Bring Claude AI Models to AWS for U.S. Government Intelligence and Defense Operations</th>\n",
+       "      <td>2.731895</td>\n",
+       "      <td>1046.315638</td>\n",
+       "      <td>2.671276</td>\n",
+       "      <td>0.000052</td>\n",
+       "      <td>15.757812</td>\n",
+       "      <td>15.361965</td>\n",
+       "      <td>2.434658</td>\n",
+       "      <td>934.908818</td>\n",
+       "      <td>2.503916</td>\n",
+       "      <td>0.000028</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Anthropic, Palantir (PLTR), AWS partnership an...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>384</td>\n",
+       "      <td>../samples/2024_anthropic_palintir.md</td>\n",
+       "      <td>Anthropic and Palantir Technologies Inc. (NYSE...</td>\n",
+       "      <td>https://investors.palantir.com/news-details/20...</td>\n",
+       "      <td>0.200000</td>\n",
+       "      <td>2024-07-11 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>TradingAgents: Multi-Agents LLM Financial Trading Framework</th>\n",
+       "      <td>3.018628</td>\n",
+       "      <td>1062.557114</td>\n",
+       "      <td>2.947409</td>\n",
+       "      <td>0.000241</td>\n",
+       "      <td>15.257812</td>\n",
+       "      <td>20.463200</td>\n",
+       "      <td>2.787077</td>\n",
+       "      <td>983.838112</td>\n",
+       "      <td>2.807337</td>\n",
+       "      <td>0.000211</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Academic paper titled \"TradingAgents: Multi-Ag...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>353</td>\n",
+       "      <td>../samples/2024_arxiv_meh.md</td>\n",
+       "      <td>TradingAgents: Multi-Agents LLM Financial Trad...</td>\n",
+       "      <td>https://arxiv.org/html/2412.20138v1</td>\n",
+       "      <td>0.150000</td>\n",
+       "      <td>2024-12-28 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Flower Crowns and Furry Mishaps by MyPalAI</th>\n",
+       "      <td>3.176868</td>\n",
+       "      <td>1528.073428</td>\n",
+       "      <td>2.790151</td>\n",
+       "      <td>0.000415</td>\n",
+       "      <td>15.187500</td>\n",
+       "      <td>23.971553</td>\n",
+       "      <td>3.057394</td>\n",
+       "      <td>1473.663968</td>\n",
+       "      <td>2.680333</td>\n",
+       "      <td>0.000613</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Sci-fi story set in Delta Eridani. Character, ...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>482</td>\n",
+       "      <td>../samples/2024_bob_fanfic.md</td>\n",
+       "      <td>Bob, April 2222 Delta Eridani\\n\\n \\n\\nThere ar...</td>\n",
+       "      <td>https://archiveofourown.org/works/61614673/cha...</td>\n",
+       "      <td>0.300000</td>\n",
+       "      <td>2024-12-24 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Deliberative Alignment: Reasoning Enables Safer Language Models</th>\n",
+       "      <td>3.224798</td>\n",
+       "      <td>1254.446484</td>\n",
+       "      <td>3.099538</td>\n",
+       "      <td>0.000083</td>\n",
+       "      <td>16.718750</td>\n",
+       "      <td>25.148498</td>\n",
+       "      <td>3.042982</td>\n",
+       "      <td>1186.763027</td>\n",
+       "      <td>2.987466</td>\n",
+       "      <td>0.000122</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Research paper on Deliberative Alignment for s...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>390</td>\n",
+       "      <td>../samples/2024_deliberative_alignment.md</td>\n",
+       "      <td># Deliberative Alignment: Reasoning Enables Sa...</td>\n",
+       "      <td>https://arxiv.org/html/2412.16339v1</td>\n",
+       "      <td>0.600000</td>\n",
+       "      <td>2024-12-20 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>fake ai hoax paper made up by gpt-4</th>\n",
+       "      <td>2.341998</td>\n",
+       "      <td>859.513342</td>\n",
+       "      <td>2.409817</td>\n",
+       "      <td>0.000074</td>\n",
+       "      <td>23.328125</td>\n",
+       "      <td>10.402001</td>\n",
+       "      <td>2.094743</td>\n",
+       "      <td>770.865261</td>\n",
+       "      <td>2.315665</td>\n",
+       "      <td>0.000097</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Academic paper. Title: \"Deconstructing Binarie...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>368</td>\n",
+       "      <td>../samples/2024_gpt4_fake_paper.md</td>\n",
+       "      <td>Title: \"Deconstructing Binaries: Interrogating...</td>\n",
+       "      <td>None</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>2024-01-01 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Hardware Hedging Against Scaling Regime Shifts (self.mlscaling)</th>\n",
+       "      <td>3.667331</td>\n",
+       "      <td>1591.621762</td>\n",
+       "      <td>3.091328</td>\n",
+       "      <td>0.001032</td>\n",
+       "      <td>15.445312</td>\n",
+       "      <td>39.147292</td>\n",
+       "      <td>3.466212</td>\n",
+       "      <td>1507.802103</td>\n",
+       "      <td>3.064746</td>\n",
+       "      <td>0.000542</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Big tech investing heavily in GPUs for paralle...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>435</td>\n",
+       "      <td>../samples/2024_gwern_reddit.md</td>\n",
+       "      <td>Hyperscalers are investing heavily in AMD/Nvid...</td>\n",
+       "      <td>https://old.reddit.com/r/mlscaling/comments/1e...</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>2024-08-21 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>How to Focus</th>\n",
+       "      <td>1.773319</td>\n",
+       "      <td>913.259251</td>\n",
+       "      <td>2.641230</td>\n",
+       "      <td>0.000018</td>\n",
+       "      <td>14.210938</td>\n",
+       "      <td>5.890371</td>\n",
+       "      <td>1.632129</td>\n",
+       "      <td>842.178410</td>\n",
+       "      <td>2.544260</td>\n",
+       "      <td>0.000024</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Book titled \"How to Focus: A Monastic Guide fo...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>516</td>\n",
+       "      <td>../samples/2024_how_to_focus.md</td>\n",
+       "      <td># HOW TO FOCUS\\n\\n\\n\\n\\n\\nANCIENT WISDOM FOR M...</td>\n",
+       "      <td>https://github.com/wassname/stoic_markdown/blo...</td>\n",
+       "      <td>0.500000</td>\n",
+       "      <td>2024-06-01 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Deontic Explorations In \"Paying To Talk To Slaves\"</th>\n",
+       "      <td>2.736291</td>\n",
+       "      <td>1203.967986</td>\n",
+       "      <td>2.828962</td>\n",
+       "      <td>0.000119</td>\n",
+       "      <td>13.765625</td>\n",
+       "      <td>15.429648</td>\n",
+       "      <td>2.494271</td>\n",
+       "      <td>1099.973593</td>\n",
+       "      <td>2.668889</td>\n",
+       "      <td>0.000089</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>AI and human engage in a Kantian ethics dialog...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>441</td>\n",
+       "      <td>../samples/2024_lesswrong_slop.md</td>\n",
+       "      <td>HER|Jenny|🤔: I want to read a dialogue between...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/Rk2o8hjYmjENH8...</td>\n",
+       "      <td>0.100000</td>\n",
+       "      <td>2024-04-12 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>By default, capital will matter more than ever after AGI</th>\n",
+       "      <td>2.991720</td>\n",
+       "      <td>1211.646540</td>\n",
+       "      <td>2.607116</td>\n",
+       "      <td>0.000087</td>\n",
+       "      <td>15.765625</td>\n",
+       "      <td>19.919912</td>\n",
+       "      <td>2.809688</td>\n",
+       "      <td>1140.733415</td>\n",
+       "      <td>2.478602</td>\n",
+       "      <td>0.000094</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>AI replacing human labor changes production fa...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>406</td>\n",
+       "      <td>../samples/2024_lw_by-default-capital-will-mat...</td>\n",
+       "      <td>**Edited to add*****:** The main takeaway of t...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/KFFaKu27FNugCH...</td>\n",
+       "      <td>0.906388</td>\n",
+       "      <td>2024-12-30 19:47:19.838000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>The Plan - 2024 Update</th>\n",
+       "      <td>2.851865</td>\n",
+       "      <td>1545.710832</td>\n",
+       "      <td>2.832593</td>\n",
+       "      <td>0.000029</td>\n",
+       "      <td>12.625000</td>\n",
+       "      <td>17.320054</td>\n",
+       "      <td>2.756729</td>\n",
+       "      <td>1496.903983</td>\n",
+       "      <td>2.823319</td>\n",
+       "      <td>0.000033</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Blog post updating progress of a project calle...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>543</td>\n",
+       "      <td>../samples/2024_lw_the-plan-2024-update.md</td>\n",
+       "      <td>This post is a follow-up to [The Plan - 2023 V...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/kJkgXEwQtWLrpe...</td>\n",
+       "      <td>0.785170</td>\n",
+       "      <td>2024-12-31 19:29:23.013000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Amazon-backed Anthropic debuts AI agents that can do complex tasks, racing against OpenAI, Microsoft and Google</th>\n",
+       "      <td>2.523585</td>\n",
+       "      <td>1324.882031</td>\n",
+       "      <td>3.052877</td>\n",
+       "      <td>0.000031</td>\n",
+       "      <td>17.296875</td>\n",
+       "      <td>12.473231</td>\n",
+       "      <td>2.356087</td>\n",
+       "      <td>1239.301737</td>\n",
+       "      <td>2.952588</td>\n",
+       "      <td>0.000029</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>AI firm Anthropic's CEO, Dario Amodei, attends...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>526</td>\n",
+       "      <td>../samples/2024_news_anthropic.md</td>\n",
+       "      <td>![PARIS, FRANCE - MAY 22: Co-founder and CEO o...</td>\n",
+       "      <td>https://www.cnbc.com/2024/10/22/anthropic-anno...</td>\n",
+       "      <td>0.500000</td>\n",
+       "      <td>2025-01-05 05:03:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>OpenAI Email Archives from Musk v. Altman</th>\n",
+       "      <td>2.922109</td>\n",
+       "      <td>1335.403592</td>\n",
+       "      <td>2.645631</td>\n",
+       "      <td>0.000162</td>\n",
+       "      <td>18.093750</td>\n",
+       "      <td>18.580423</td>\n",
+       "      <td>2.782840</td>\n",
+       "      <td>1274.540800</td>\n",
+       "      <td>2.565053</td>\n",
+       "      <td>0.000472</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Email exchange between Sam Altman and Elon Mus...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>458</td>\n",
+       "      <td>../samples/2024_openai_emails.md</td>\n",
+       "      <td>Subject: question\\nSam Altman to Elon Musk - M...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/5jjk4CDnj9tA7u...</td>\n",
+       "      <td>0.700000</td>\n",
+       "      <td>2024-11-01 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>President Trump Announces Morgan Ortagus as Deputy Special Presidential Envoy for Middle East Peace</th>\n",
+       "      <td>2.811518</td>\n",
+       "      <td>674.764250</td>\n",
+       "      <td>2.876954</td>\n",
+       "      <td>0.000784</td>\n",
+       "      <td>20.031250</td>\n",
+       "      <td>16.635146</td>\n",
+       "      <td>2.595940</td>\n",
+       "      <td>625.621626</td>\n",
+       "      <td>2.555692</td>\n",
+       "      <td>0.000939</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Morgan Ortagus named Deputy Special Pres. Envo...</td>\n",
+       "      <td>1061</td>\n",
+       "      <td>241</td>\n",
+       "      <td>../samples/2024_trump_appointment.md</td>\n",
+       "      <td>I am pleased to announce Morgan Ortagus as Dep...</td>\n",
+       "      <td>https://truthsocial.com/@realDonaldTrump/posts...</td>\n",
+       "      <td>0.300000</td>\n",
+       "      <td>2025-01-04 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>CDC Report on Missouri H5N1 Serology Testing</th>\n",
+       "      <td>2.616612</td>\n",
+       "      <td>986.462712</td>\n",
+       "      <td>2.657349</td>\n",
+       "      <td>0.000595</td>\n",
+       "      <td>14.695312</td>\n",
+       "      <td>13.689265</td>\n",
+       "      <td>2.427171</td>\n",
+       "      <td>917.470712</td>\n",
+       "      <td>2.531456</td>\n",
+       "      <td>0.000496</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>CDC report, Oct 24, 2024. Missouri patient, fl...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>378</td>\n",
+       "      <td>../samples/2025_h5n1_report.md</td>\n",
+       "      <td>## CDC Report on Missouri H5N1 Serology Testin...</td>\n",
+       "      <td>https://arxiv.org/html/2412.16339v1</td>\n",
+       "      <td>0.750000</td>\n",
+       "      <td>2025-01-05 00:00:00+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2024 in AI predictions</th>\n",
+       "      <td>3.666700</td>\n",
+       "      <td>1914.017552</td>\n",
+       "      <td>3.061472</td>\n",
+       "      <td>0.000343</td>\n",
+       "      <td>13.562500</td>\n",
+       "      <td>39.122599</td>\n",
+       "      <td>3.569884</td>\n",
+       "      <td>1867.049485</td>\n",
+       "      <td>3.025535</td>\n",
+       "      <td>0.000414</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>AI prediction review 2024 blog post. Casual-to...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>523</td>\n",
+       "      <td>../samples/2025_lw_2024-in-ai-predictions.md</td>\n",
+       "      <td>Follow-up to: [2023 in AI predictions](https:/...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/CJ4sppkGcbnGMS...</td>\n",
+       "      <td>0.789190</td>\n",
+       "      <td>2025-01-02 02:09:13.093000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Comment on 'Death and the Gorgon'</th>\n",
+       "      <td>3.371871</td>\n",
+       "      <td>1537.573345</td>\n",
+       "      <td>3.256970</td>\n",
+       "      <td>0.000051</td>\n",
+       "      <td>15.078125</td>\n",
+       "      <td>29.132995</td>\n",
+       "      <td>3.177145</td>\n",
+       "      <td>1451.955038</td>\n",
+       "      <td>3.199762</td>\n",
+       "      <td>0.000098</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Book review, Greg Egan, contrast between works...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>457</td>\n",
+       "      <td>../samples/2025_lw_comment-on-death-and-the-go...</td>\n",
+       "      <td>*(some plot spoilers)*\\n\\nThere's something di...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/hx5EkHFH5hGzng...</td>\n",
+       "      <td>0.750253</td>\n",
+       "      <td>2025-01-01 06:00:24.498000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>debating buying NVDA in 2019</th>\n",
+       "      <td>2.933237</td>\n",
+       "      <td>1308.223717</td>\n",
+       "      <td>2.733556</td>\n",
+       "      <td>0.000224</td>\n",
+       "      <td>12.953125</td>\n",
+       "      <td>18.788351</td>\n",
+       "      <td>2.784547</td>\n",
+       "      <td>1244.692610</td>\n",
+       "      <td>2.696586</td>\n",
+       "      <td>0.000320</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Dialogue, tech-related discussion. Alice, upbe...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>447</td>\n",
+       "      <td>../samples/2025_lw_debating-buying-nvda-in-201...</td>\n",
+       "      <td>Alice: You saw GPT-2, right?\\n\\nBob: Of course...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/QFSZmMdtPwoAHd...</td>\n",
+       "      <td>0.526975</td>\n",
+       "      <td>2025-01-04 18:07:35.832000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Human study on AI spear phishing campaigns</th>\n",
+       "      <td>2.849588</td>\n",
+       "      <td>1199.676460</td>\n",
+       "      <td>2.964267</td>\n",
+       "      <td>0.000076</td>\n",
+       "      <td>22.140625</td>\n",
+       "      <td>17.280657</td>\n",
+       "      <td>2.527051</td>\n",
+       "      <td>1066.415470</td>\n",
+       "      <td>2.778351</td>\n",
+       "      <td>0.000067</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Study: human subjects, AI spear-phishing effec...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>422</td>\n",
+       "      <td>../samples/2025_lw_human-study-on-ai-spear-phi...</td>\n",
+       "      <td>TL;DR: We ran a human subject study on whether...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/GCHyDKfPXa5qsG...</td>\n",
+       "      <td>0.677458</td>\n",
+       "      <td>2025-01-03 19:03:28.406000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>My AGI safety research—2024 review, ’25 plans</th>\n",
+       "      <td>2.705293</td>\n",
+       "      <td>1355.351999</td>\n",
+       "      <td>2.934107</td>\n",
+       "      <td>0.000060</td>\n",
+       "      <td>13.671875</td>\n",
+       "      <td>14.958705</td>\n",
+       "      <td>2.539809</td>\n",
+       "      <td>1274.984361</td>\n",
+       "      <td>2.829353</td>\n",
+       "      <td>0.000105</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Research review &amp; plans doc. Style explanatory...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>502</td>\n",
+       "      <td>../samples/2025_lw_my-agi-safety-research-2024...</td>\n",
+       "      <td>*Previous:*[*My AGI safety research—2022 revie...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/2wHaCimHehsF36...</td>\n",
+       "      <td>0.756925</td>\n",
+       "      <td>2025-01-01 22:01:48.820000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Parkinson's Law and the Ideology of Statistics</th>\n",
+       "      <td>3.215346</td>\n",
+       "      <td>1231.477661</td>\n",
+       "      <td>3.011060</td>\n",
+       "      <td>0.000046</td>\n",
+       "      <td>14.421875</td>\n",
+       "      <td>24.911919</td>\n",
+       "      <td>2.927131</td>\n",
+       "      <td>1124.018382</td>\n",
+       "      <td>2.878317</td>\n",
+       "      <td>0.000056</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Book review of *The Anti-Politics Machine* on ...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>384</td>\n",
+       "      <td>../samples/2025_lw_parkinson-s-law-and-the-ide...</td>\n",
+       "      <td>The [anonymous review of *The Anti-Politics Ma...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/4CmYSPc4HfRfWx...</td>\n",
+       "      <td>0.677458</td>\n",
+       "      <td>2025-01-04 22:59:57.376000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Preference Inversion</th>\n",
+       "      <td>3.366223</td>\n",
+       "      <td>1380.151608</td>\n",
+       "      <td>2.890657</td>\n",
+       "      <td>0.000076</td>\n",
+       "      <td>13.054688</td>\n",
+       "      <td>28.968917</td>\n",
+       "      <td>3.222520</td>\n",
+       "      <td>1324.455591</td>\n",
+       "      <td>2.831317</td>\n",
+       "      <td>0.000298</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Text discusses how reported personal preferenc...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>411</td>\n",
+       "      <td>../samples/2025_lw_preference-inversion.md</td>\n",
+       "      <td>Sometimes the preferences people report or eve...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/zMkQFuNqMBpBvu...</td>\n",
+       "      <td>0.633321</td>\n",
+       "      <td>2025-01-03 23:49:06.168000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Review: Planecrash</th>\n",
+       "      <td>3.105758</td>\n",
+       "      <td>1844.820265</td>\n",
+       "      <td>3.259858</td>\n",
+       "      <td>0.000006</td>\n",
+       "      <td>15.101562</td>\n",
+       "      <td>22.326136</td>\n",
+       "      <td>2.891863</td>\n",
+       "      <td>1720.658635</td>\n",
+       "      <td>3.165378</td>\n",
+       "      <td>0.000006</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Review of \"Planecrash\", a fiction board-style ...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>595</td>\n",
+       "      <td>../samples/2025_lw_review-planecrash.md</td>\n",
+       "      <td>Take a stereotypical fantasy novel, a textbook...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/zRHGQ9f6deKbxJ...</td>\n",
+       "      <td>0.933950</td>\n",
+       "      <td>2025-01-01 01:31:26.864000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>The Field of AI Alignment: A Postmortem, and What To Do About It</th>\n",
+       "      <td>2.901739</td>\n",
+       "      <td>1178.106149</td>\n",
+       "      <td>2.826933</td>\n",
+       "      <td>0.000158</td>\n",
+       "      <td>14.648438</td>\n",
+       "      <td>18.205783</td>\n",
+       "      <td>2.797740</td>\n",
+       "      <td>1138.680238</td>\n",
+       "      <td>2.806392</td>\n",
+       "      <td>0.000153</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>AI alignment research critique; Author disappo...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>407</td>\n",
+       "      <td>../samples/2025_lw_the-field-of-ai-alignment-a...</td>\n",
+       "      <td>&gt; A policeman sees a drunk man searching for s...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/nwpyhyagpPYDn4...</td>\n",
+       "      <td>0.925483</td>\n",
+       "      <td>2025-01-01 00:42:36.538000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>The Intelligence Curse</th>\n",
+       "      <td>3.326115</td>\n",
+       "      <td>1377.011554</td>\n",
+       "      <td>2.957692</td>\n",
+       "      <td>0.000340</td>\n",
+       "      <td>15.570312</td>\n",
+       "      <td>27.830008</td>\n",
+       "      <td>3.212695</td>\n",
+       "      <td>1333.268286</td>\n",
+       "      <td>2.888920</td>\n",
+       "      <td>0.000819</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Economical look at AGI, comparison: AGI to res...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>415</td>\n",
+       "      <td>../samples/2025_lw_the-intelligence-curse.md</td>\n",
+       "      <td>*“Show me the incentive, and I’ll show you the...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/Mak2kZuTq8Hpnq...</td>\n",
+       "      <td>0.688044</td>\n",
+       "      <td>2025-01-04 18:16:58.921000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>The Laws of Large Numbers</th>\n",
+       "      <td>2.630560</td>\n",
+       "      <td>1165.338041</td>\n",
+       "      <td>2.688951</td>\n",
+       "      <td>0.000065</td>\n",
+       "      <td>13.429688</td>\n",
+       "      <td>13.881540</td>\n",
+       "      <td>2.505327</td>\n",
+       "      <td>1112.365011</td>\n",
+       "      <td>2.579544</td>\n",
+       "      <td>0.000062</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Blog post discussing advanced law of large num...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>444</td>\n",
+       "      <td>../samples/2025_lw_the-laws-of-large-numbers.md</td>\n",
+       "      <td>Introduction\\n============\\n\\nIn this short po...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/EhTMM77iKBTBxB...</td>\n",
+       "      <td>0.540932</td>\n",
+       "      <td>2025-01-04 18:06:02.387000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>The subset parity learning problem: much more than you wanted to know</th>\n",
+       "      <td>2.688764</td>\n",
+       "      <td>1150.790915</td>\n",
+       "      <td>2.577458</td>\n",
+       "      <td>0.000050</td>\n",
+       "      <td>12.570312</td>\n",
+       "      <td>14.713476</td>\n",
+       "      <td>2.549593</td>\n",
+       "      <td>1093.775236</td>\n",
+       "      <td>2.426958</td>\n",
+       "      <td>0.000058</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Blog post discussing parity learning problem i...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>429</td>\n",
+       "      <td>../samples/2025_lw_the-subset-parity-learning-...</td>\n",
+       "      <td>Imagine that you’re looking for buried treasur...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/Mcrfi3DBJBzfoL...</td>\n",
+       "      <td>0.697946</td>\n",
+       "      <td>2025-01-04 09:59:16.158000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>What’s the short timeline plan?</th>\n",
+       "      <td>3.330626</td>\n",
+       "      <td>1575.386132</td>\n",
+       "      <td>2.826830</td>\n",
+       "      <td>0.000362</td>\n",
+       "      <td>13.468750</td>\n",
+       "      <td>27.955839</td>\n",
+       "      <td>3.037546</td>\n",
+       "      <td>1439.797010</td>\n",
+       "      <td>2.714645</td>\n",
+       "      <td>0.000317</td>\n",
+       "      <td>...</td>\n",
+       "      <td>unsloth/Llama-3.2-1B-bnb-4bit</td>\n",
+       "      <td>Opinion blog post by Apollo Research member ex...</td>\n",
+       "      <td>2000</td>\n",
+       "      <td>474</td>\n",
+       "      <td>../samples/2025_lw_what-s-the-short-timeline-p...</td>\n",
+       "      <td>*This is a low-effort post. I mostly want to g...</td>\n",
+       "      <td>https://www.lesswrong.com/posts/bb5Tnjdrptu89r...</td>\n",
+       "      <td>0.898161</td>\n",
+       "      <td>2025-01-05 00:10:28.708000+00:00</td>\n",
+       "      <td>False</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>27 rows × 22 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                                                    before_mean   before_sum  \\\n",
+       "title                                                                          \n",
+       "Anthropic and Palantir Partner to Bring Claude ...     2.731895  1046.315638   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...     3.018628  1062.557114   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI             3.176868  1528.073428   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...     3.224798  1254.446484   \n",
+       "fake ai hoax paper made up by gpt-4                    2.341998   859.513342   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...     3.667331  1591.621762   \n",
+       "How to Focus                                           1.773319   913.259251   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"     2.736291  1203.967986   \n",
+       "By default, capital will matter more than ever ...     2.991720  1211.646540   \n",
+       "The Plan - 2024 Update                                 2.851865  1545.710832   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...     2.523585  1324.882031   \n",
+       "OpenAI Email Archives from Musk v. Altman              2.922109  1335.403592   \n",
+       "President Trump Announces Morgan Ortagus as Dep...     2.811518   674.764250   \n",
+       "CDC Report on Missouri H5N1 Serology Testing           2.616612   986.462712   \n",
+       "2024 in AI predictions                                 3.666700  1914.017552   \n",
+       "Comment on 'Death and the Gorgon'                      3.371871  1537.573345   \n",
+       "debating buying NVDA in 2019                           2.933237  1308.223717   \n",
+       "Human study on AI spear phishing campaigns             2.849588  1199.676460   \n",
+       "My AGI safety research—2024 review, ’25 plans          2.705293  1355.351999   \n",
+       "Parkinson's Law and the Ideology of Statistics         3.215346  1231.477661   \n",
+       "Preference Inversion                                   3.366223  1380.151608   \n",
+       "Review: Planecrash                                     3.105758  1844.820265   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...     2.901739  1178.106149   \n",
+       "The Intelligence Curse                                 3.326115  1377.011554   \n",
+       "The Laws of Large Numbers                              2.630560  1165.338041   \n",
+       "The subset parity learning problem: much more t...     2.688764  1150.790915   \n",
+       "What’s the short timeline plan?                        3.330626  1575.386132   \n",
+       "\n",
+       "                                                    before_std  before_min  \\\n",
+       "title                                                                        \n",
+       "Anthropic and Palantir Partner to Bring Claude ...    2.671276    0.000052   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...    2.947409    0.000241   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI            2.790151    0.000415   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...    3.099538    0.000083   \n",
+       "fake ai hoax paper made up by gpt-4                   2.409817    0.000074   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...    3.091328    0.001032   \n",
+       "How to Focus                                          2.641230    0.000018   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"    2.828962    0.000119   \n",
+       "By default, capital will matter more than ever ...    2.607116    0.000087   \n",
+       "The Plan - 2024 Update                                2.832593    0.000029   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...    3.052877    0.000031   \n",
+       "OpenAI Email Archives from Musk v. Altman             2.645631    0.000162   \n",
+       "President Trump Announces Morgan Ortagus as Dep...    2.876954    0.000784   \n",
+       "CDC Report on Missouri H5N1 Serology Testing          2.657349    0.000595   \n",
+       "2024 in AI predictions                                3.061472    0.000343   \n",
+       "Comment on 'Death and the Gorgon'                     3.256970    0.000051   \n",
+       "debating buying NVDA in 2019                          2.733556    0.000224   \n",
+       "Human study on AI spear phishing campaigns            2.964267    0.000076   \n",
+       "My AGI safety research—2024 review, ’25 plans         2.934107    0.000060   \n",
+       "Parkinson's Law and the Ideology of Statistics        3.011060    0.000046   \n",
+       "Preference Inversion                                  2.890657    0.000076   \n",
+       "Review: Planecrash                                    3.259858    0.000006   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...    2.826933    0.000158   \n",
+       "The Intelligence Curse                                2.957692    0.000340   \n",
+       "The Laws of Large Numbers                             2.688951    0.000065   \n",
+       "The subset parity learning problem: much more t...    2.577458    0.000050   \n",
+       "What’s the short timeline plan?                       2.826830    0.000362   \n",
+       "\n",
+       "                                                    before_max  before_ppx  \\\n",
+       "title                                                                        \n",
+       "Anthropic and Palantir Partner to Bring Claude ...   15.757812   15.361965   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...   15.257812   20.463200   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI           15.187500   23.971553   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...   16.718750   25.148498   \n",
+       "fake ai hoax paper made up by gpt-4                  23.328125   10.402001   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...   15.445312   39.147292   \n",
+       "How to Focus                                         14.210938    5.890371   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"   13.765625   15.429648   \n",
+       "By default, capital will matter more than ever ...   15.765625   19.919912   \n",
+       "The Plan - 2024 Update                               12.625000   17.320054   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...   17.296875   12.473231   \n",
+       "OpenAI Email Archives from Musk v. Altman            18.093750   18.580423   \n",
+       "President Trump Announces Morgan Ortagus as Dep...   20.031250   16.635146   \n",
+       "CDC Report on Missouri H5N1 Serology Testing         14.695312   13.689265   \n",
+       "2024 in AI predictions                               13.562500   39.122599   \n",
+       "Comment on 'Death and the Gorgon'                    15.078125   29.132995   \n",
+       "debating buying NVDA in 2019                         12.953125   18.788351   \n",
+       "Human study on AI spear phishing campaigns           22.140625   17.280657   \n",
+       "My AGI safety research—2024 review, ’25 plans        13.671875   14.958705   \n",
+       "Parkinson's Law and the Ideology of Statistics       14.421875   24.911919   \n",
+       "Preference Inversion                                 13.054688   28.968917   \n",
+       "Review: Planecrash                                   15.101562   22.326136   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...   14.648438   18.205783   \n",
+       "The Intelligence Curse                               15.570312   27.830008   \n",
+       "The Laws of Large Numbers                            13.429688   13.881540   \n",
+       "The subset parity learning problem: much more t...   12.570312   14.713476   \n",
+       "What’s the short timeline plan?                      13.468750   27.955839   \n",
+       "\n",
+       "                                                    after_mean    after_sum  \\\n",
+       "title                                                                         \n",
+       "Anthropic and Palantir Partner to Bring Claude ...    2.434658   934.908818   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...    2.787077   983.838112   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI            3.057394  1473.663968   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...    3.042982  1186.763027   \n",
+       "fake ai hoax paper made up by gpt-4                   2.094743   770.865261   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...    3.466212  1507.802103   \n",
+       "How to Focus                                          1.632129   842.178410   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"    2.494271  1099.973593   \n",
+       "By default, capital will matter more than ever ...    2.809688  1140.733415   \n",
+       "The Plan - 2024 Update                                2.756729  1496.903983   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...    2.356087  1239.301737   \n",
+       "OpenAI Email Archives from Musk v. Altman             2.782840  1274.540800   \n",
+       "President Trump Announces Morgan Ortagus as Dep...    2.595940   625.621626   \n",
+       "CDC Report on Missouri H5N1 Serology Testing          2.427171   917.470712   \n",
+       "2024 in AI predictions                                3.569884  1867.049485   \n",
+       "Comment on 'Death and the Gorgon'                     3.177145  1451.955038   \n",
+       "debating buying NVDA in 2019                          2.784547  1244.692610   \n",
+       "Human study on AI spear phishing campaigns            2.527051  1066.415470   \n",
+       "My AGI safety research—2024 review, ’25 plans         2.539809  1274.984361   \n",
+       "Parkinson's Law and the Ideology of Statistics        2.927131  1124.018382   \n",
+       "Preference Inversion                                  3.222520  1324.455591   \n",
+       "Review: Planecrash                                    2.891863  1720.658635   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...    2.797740  1138.680238   \n",
+       "The Intelligence Curse                                3.212695  1333.268286   \n",
+       "The Laws of Large Numbers                             2.505327  1112.365011   \n",
+       "The subset parity learning problem: much more t...    2.549593  1093.775236   \n",
+       "What’s the short timeline plan?                       3.037546  1439.797010   \n",
+       "\n",
+       "                                                    after_std  after_min  ...  \\\n",
+       "title                                                                     ...   \n",
+       "Anthropic and Palantir Partner to Bring Claude ...   2.503916   0.000028  ...   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...   2.807337   0.000211  ...   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI           2.680333   0.000613  ...   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...   2.987466   0.000122  ...   \n",
+       "fake ai hoax paper made up by gpt-4                  2.315665   0.000097  ...   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...   3.064746   0.000542  ...   \n",
+       "How to Focus                                         2.544260   0.000024  ...   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"   2.668889   0.000089  ...   \n",
+       "By default, capital will matter more than ever ...   2.478602   0.000094  ...   \n",
+       "The Plan - 2024 Update                               2.823319   0.000033  ...   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...   2.952588   0.000029  ...   \n",
+       "OpenAI Email Archives from Musk v. Altman            2.565053   0.000472  ...   \n",
+       "President Trump Announces Morgan Ortagus as Dep...   2.555692   0.000939  ...   \n",
+       "CDC Report on Missouri H5N1 Serology Testing         2.531456   0.000496  ...   \n",
+       "2024 in AI predictions                               3.025535   0.000414  ...   \n",
+       "Comment on 'Death and the Gorgon'                    3.199762   0.000098  ...   \n",
+       "debating buying NVDA in 2019                         2.696586   0.000320  ...   \n",
+       "Human study on AI spear phishing campaigns           2.778351   0.000067  ...   \n",
+       "My AGI safety research—2024 review, ’25 plans        2.829353   0.000105  ...   \n",
+       "Parkinson's Law and the Ideology of Statistics       2.878317   0.000056  ...   \n",
+       "Preference Inversion                                 2.831317   0.000298  ...   \n",
+       "Review: Planecrash                                   3.165378   0.000006  ...   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...   2.806392   0.000153  ...   \n",
+       "The Intelligence Curse                               2.888920   0.000819  ...   \n",
+       "The Laws of Large Numbers                            2.579544   0.000062  ...   \n",
+       "The subset parity learning problem: much more t...   2.426958   0.000058  ...   \n",
+       "What’s the short timeline plan?                      2.714645   0.000317  ...   \n",
+       "\n",
+       "                                                                            model  \\\n",
+       "title                                                                               \n",
+       "Anthropic and Palantir Partner to Bring Claude ...  unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...  unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI          unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...  unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "fake ai hoax paper made up by gpt-4                 unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...  unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "How to Focus                                        unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"  unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "By default, capital will matter more than ever ...  unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "The Plan - 2024 Update                              unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...  unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "OpenAI Email Archives from Musk v. Altman           unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "President Trump Announces Morgan Ortagus as Dep...  unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "CDC Report on Missouri H5N1 Serology Testing        unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "2024 in AI predictions                              unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "Comment on 'Death and the Gorgon'                   unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "debating buying NVDA in 2019                        unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "Human study on AI spear phishing campaigns          unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "My AGI safety research—2024 review, ’25 plans       unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "Parkinson's Law and the Ideology of Statistics      unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "Preference Inversion                                unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "Review: Planecrash                                  unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...  unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "The Intelligence Curse                              unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "The Laws of Large Numbers                           unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "The subset parity learning problem: much more t...  unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "What’s the short timeline plan?                     unsloth/Llama-3.2-1B-bnb-4bit   \n",
+       "\n",
+       "                                                                                              summary  \\\n",
+       "title                                                                                                   \n",
+       "Anthropic and Palantir Partner to Bring Claude ...  Anthropic, Palantir (PLTR), AWS partnership an...   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...  Academic paper titled \"TradingAgents: Multi-Ag...   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI          Sci-fi story set in Delta Eridani. Character, ...   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...  Research paper on Deliberative Alignment for s...   \n",
+       "fake ai hoax paper made up by gpt-4                 Academic paper. Title: \"Deconstructing Binarie...   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...  Big tech investing heavily in GPUs for paralle...   \n",
+       "How to Focus                                        Book titled \"How to Focus: A Monastic Guide fo...   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"  AI and human engage in a Kantian ethics dialog...   \n",
+       "By default, capital will matter more than ever ...  AI replacing human labor changes production fa...   \n",
+       "The Plan - 2024 Update                              Blog post updating progress of a project calle...   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...  AI firm Anthropic's CEO, Dario Amodei, attends...   \n",
+       "OpenAI Email Archives from Musk v. Altman           Email exchange between Sam Altman and Elon Mus...   \n",
+       "President Trump Announces Morgan Ortagus as Dep...  Morgan Ortagus named Deputy Special Pres. Envo...   \n",
+       "CDC Report on Missouri H5N1 Serology Testing        CDC report, Oct 24, 2024. Missouri patient, fl...   \n",
+       "2024 in AI predictions                              AI prediction review 2024 blog post. Casual-to...   \n",
+       "Comment on 'Death and the Gorgon'                   Book review, Greg Egan, contrast between works...   \n",
+       "debating buying NVDA in 2019                        Dialogue, tech-related discussion. Alice, upbe...   \n",
+       "Human study on AI spear phishing campaigns          Study: human subjects, AI spear-phishing effec...   \n",
+       "My AGI safety research—2024 review, ’25 plans       Research review & plans doc. Style explanatory...   \n",
+       "Parkinson's Law and the Ideology of Statistics      Book review of *The Anti-Politics Machine* on ...   \n",
+       "Preference Inversion                                Text discusses how reported personal preferenc...   \n",
+       "Review: Planecrash                                  Review of \"Planecrash\", a fiction board-style ...   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...  AI alignment research critique; Author disappo...   \n",
+       "The Intelligence Curse                              Economical look at AGI, comparison: AGI to res...   \n",
+       "The Laws of Large Numbers                           Blog post discussing advanced law of large num...   \n",
+       "The subset parity learning problem: much more t...  Blog post discussing parity learning problem i...   \n",
+       "What’s the short timeline plan?                     Opinion blog post by Apollo Research member ex...   \n",
+       "\n",
+       "                                                     len n_tokens  \\\n",
+       "title                                                               \n",
+       "Anthropic and Palantir Partner to Bring Claude ...  2000      384   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...  2000      353   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI          2000      482   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...  2000      390   \n",
+       "fake ai hoax paper made up by gpt-4                 2000      368   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...  2000      435   \n",
+       "How to Focus                                        2000      516   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"  2000      441   \n",
+       "By default, capital will matter more than ever ...  2000      406   \n",
+       "The Plan - 2024 Update                              2000      543   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...  2000      526   \n",
+       "OpenAI Email Archives from Musk v. Altman           2000      458   \n",
+       "President Trump Announces Morgan Ortagus as Dep...  1061      241   \n",
+       "CDC Report on Missouri H5N1 Serology Testing        2000      378   \n",
+       "2024 in AI predictions                              2000      523   \n",
+       "Comment on 'Death and the Gorgon'                   2000      457   \n",
+       "debating buying NVDA in 2019                        2000      447   \n",
+       "Human study on AI spear phishing campaigns          2000      422   \n",
+       "My AGI safety research—2024 review, ’25 plans       2000      502   \n",
+       "Parkinson's Law and the Ideology of Statistics      2000      384   \n",
+       "Preference Inversion                                2000      411   \n",
+       "Review: Planecrash                                  2000      595   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...  2000      407   \n",
+       "The Intelligence Curse                              2000      415   \n",
+       "The Laws of Large Numbers                           2000      444   \n",
+       "The subset parity learning problem: much more t...  2000      429   \n",
+       "What’s the short timeline plan?                     2000      474   \n",
+       "\n",
+       "                                                                                                    f  \\\n",
+       "title                                                                                                   \n",
+       "Anthropic and Palantir Partner to Bring Claude ...              ../samples/2024_anthropic_palintir.md   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...                       ../samples/2024_arxiv_meh.md   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI                              ../samples/2024_bob_fanfic.md   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...          ../samples/2024_deliberative_alignment.md   \n",
+       "fake ai hoax paper made up by gpt-4                                ../samples/2024_gpt4_fake_paper.md   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...                    ../samples/2024_gwern_reddit.md   \n",
+       "How to Focus                                                          ../samples/2024_how_to_focus.md   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"                  ../samples/2024_lesswrong_slop.md   \n",
+       "By default, capital will matter more than ever ...  ../samples/2024_lw_by-default-capital-will-mat...   \n",
+       "The Plan - 2024 Update                                     ../samples/2024_lw_the-plan-2024-update.md   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...                  ../samples/2024_news_anthropic.md   \n",
+       "OpenAI Email Archives from Musk v. Altman                            ../samples/2024_openai_emails.md   \n",
+       "President Trump Announces Morgan Ortagus as Dep...               ../samples/2024_trump_appointment.md   \n",
+       "CDC Report on Missouri H5N1 Serology Testing                           ../samples/2025_h5n1_report.md   \n",
+       "2024 in AI predictions                                   ../samples/2025_lw_2024-in-ai-predictions.md   \n",
+       "Comment on 'Death and the Gorgon'                   ../samples/2025_lw_comment-on-death-and-the-go...   \n",
+       "debating buying NVDA in 2019                        ../samples/2025_lw_debating-buying-nvda-in-201...   \n",
+       "Human study on AI spear phishing campaigns          ../samples/2025_lw_human-study-on-ai-spear-phi...   \n",
+       "My AGI safety research—2024 review, ’25 plans       ../samples/2025_lw_my-agi-safety-research-2024...   \n",
+       "Parkinson's Law and the Ideology of Statistics      ../samples/2025_lw_parkinson-s-law-and-the-ide...   \n",
+       "Preference Inversion                                       ../samples/2025_lw_preference-inversion.md   \n",
+       "Review: Planecrash                                            ../samples/2025_lw_review-planecrash.md   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...  ../samples/2025_lw_the-field-of-ai-alignment-a...   \n",
+       "The Intelligence Curse                                   ../samples/2025_lw_the-intelligence-curse.md   \n",
+       "The Laws of Large Numbers                             ../samples/2025_lw_the-laws-of-large-numbers.md   \n",
+       "The subset parity learning problem: much more t...  ../samples/2025_lw_the-subset-parity-learning-...   \n",
+       "What’s the short timeline plan?                     ../samples/2025_lw_what-s-the-short-timeline-p...   \n",
+       "\n",
+       "                                                                                              content  \\\n",
+       "title                                                                                                   \n",
+       "Anthropic and Palantir Partner to Bring Claude ...  Anthropic and Palantir Technologies Inc. (NYSE...   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...  TradingAgents: Multi-Agents LLM Financial Trad...   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI          Bob, April 2222 Delta Eridani\\n\\n \\n\\nThere ar...   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...  # Deliberative Alignment: Reasoning Enables Sa...   \n",
+       "fake ai hoax paper made up by gpt-4                 Title: \"Deconstructing Binaries: Interrogating...   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...  Hyperscalers are investing heavily in AMD/Nvid...   \n",
+       "How to Focus                                        # HOW TO FOCUS\\n\\n\\n\\n\\n\\nANCIENT WISDOM FOR M...   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"  HER|Jenny|🤔: I want to read a dialogue between...   \n",
+       "By default, capital will matter more than ever ...  **Edited to add*****:** The main takeaway of t...   \n",
+       "The Plan - 2024 Update                              This post is a follow-up to [The Plan - 2023 V...   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...  ![PARIS, FRANCE - MAY 22: Co-founder and CEO o...   \n",
+       "OpenAI Email Archives from Musk v. Altman           Subject: question\\nSam Altman to Elon Musk - M...   \n",
+       "President Trump Announces Morgan Ortagus as Dep...  I am pleased to announce Morgan Ortagus as Dep...   \n",
+       "CDC Report on Missouri H5N1 Serology Testing        ## CDC Report on Missouri H5N1 Serology Testin...   \n",
+       "2024 in AI predictions                              Follow-up to: [2023 in AI predictions](https:/...   \n",
+       "Comment on 'Death and the Gorgon'                   *(some plot spoilers)*\\n\\nThere's something di...   \n",
+       "debating buying NVDA in 2019                        Alice: You saw GPT-2, right?\\n\\nBob: Of course...   \n",
+       "Human study on AI spear phishing campaigns          TL;DR: We ran a human subject study on whether...   \n",
+       "My AGI safety research—2024 review, ’25 plans       *Previous:*[*My AGI safety research—2022 revie...   \n",
+       "Parkinson's Law and the Ideology of Statistics      The [anonymous review of *The Anti-Politics Ma...   \n",
+       "Preference Inversion                                Sometimes the preferences people report or eve...   \n",
+       "Review: Planecrash                                  Take a stereotypical fantasy novel, a textbook...   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...  > A policeman sees a drunk man searching for s...   \n",
+       "The Intelligence Curse                              *“Show me the incentive, and I’ll show you the...   \n",
+       "The Laws of Large Numbers                           Introduction\\n============\\n\\nIn this short po...   \n",
+       "The subset parity learning problem: much more t...  Imagine that you’re looking for buried treasur...   \n",
+       "What’s the short timeline plan?                     *This is a low-effort post. I mostly want to g...   \n",
+       "\n",
+       "                                                                                                  url  \\\n",
+       "title                                                                                                   \n",
+       "Anthropic and Palantir Partner to Bring Claude ...  https://investors.palantir.com/news-details/20...   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...                https://arxiv.org/html/2412.20138v1   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI          https://archiveofourown.org/works/61614673/cha...   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...                https://arxiv.org/html/2412.16339v1   \n",
+       "fake ai hoax paper made up by gpt-4                                                              None   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...  https://old.reddit.com/r/mlscaling/comments/1e...   \n",
+       "How to Focus                                        https://github.com/wassname/stoic_markdown/blo...   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"  https://www.lesswrong.com/posts/Rk2o8hjYmjENH8...   \n",
+       "By default, capital will matter more than ever ...  https://www.lesswrong.com/posts/KFFaKu27FNugCH...   \n",
+       "The Plan - 2024 Update                              https://www.lesswrong.com/posts/kJkgXEwQtWLrpe...   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...  https://www.cnbc.com/2024/10/22/anthropic-anno...   \n",
+       "OpenAI Email Archives from Musk v. Altman           https://www.lesswrong.com/posts/5jjk4CDnj9tA7u...   \n",
+       "President Trump Announces Morgan Ortagus as Dep...  https://truthsocial.com/@realDonaldTrump/posts...   \n",
+       "CDC Report on Missouri H5N1 Serology Testing                      https://arxiv.org/html/2412.16339v1   \n",
+       "2024 in AI predictions                              https://www.lesswrong.com/posts/CJ4sppkGcbnGMS...   \n",
+       "Comment on 'Death and the Gorgon'                   https://www.lesswrong.com/posts/hx5EkHFH5hGzng...   \n",
+       "debating buying NVDA in 2019                        https://www.lesswrong.com/posts/QFSZmMdtPwoAHd...   \n",
+       "Human study on AI spear phishing campaigns          https://www.lesswrong.com/posts/GCHyDKfPXa5qsG...   \n",
+       "My AGI safety research—2024 review, ’25 plans       https://www.lesswrong.com/posts/2wHaCimHehsF36...   \n",
+       "Parkinson's Law and the Ideology of Statistics      https://www.lesswrong.com/posts/4CmYSPc4HfRfWx...   \n",
+       "Preference Inversion                                https://www.lesswrong.com/posts/zMkQFuNqMBpBvu...   \n",
+       "Review: Planecrash                                  https://www.lesswrong.com/posts/zRHGQ9f6deKbxJ...   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...  https://www.lesswrong.com/posts/nwpyhyagpPYDn4...   \n",
+       "The Intelligence Curse                              https://www.lesswrong.com/posts/Mak2kZuTq8Hpnq...   \n",
+       "The Laws of Large Numbers                           https://www.lesswrong.com/posts/EhTMM77iKBTBxB...   \n",
+       "The subset parity learning problem: much more t...  https://www.lesswrong.com/posts/Mcrfi3DBJBzfoL...   \n",
+       "What’s the short timeline plan?                     https://www.lesswrong.com/posts/bb5Tnjdrptu89r...   \n",
+       "\n",
+       "                                                     novelty  \\\n",
+       "title                                                          \n",
+       "Anthropic and Palantir Partner to Bring Claude ...  0.200000   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...  0.150000   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI          0.300000   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...  0.600000   \n",
+       "fake ai hoax paper made up by gpt-4                 0.000000   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...  1.000000   \n",
+       "How to Focus                                        0.500000   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"  0.100000   \n",
+       "By default, capital will matter more than ever ...  0.906388   \n",
+       "The Plan - 2024 Update                              0.785170   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...  0.500000   \n",
+       "OpenAI Email Archives from Musk v. Altman           0.700000   \n",
+       "President Trump Announces Morgan Ortagus as Dep...  0.300000   \n",
+       "CDC Report on Missouri H5N1 Serology Testing        0.750000   \n",
+       "2024 in AI predictions                              0.789190   \n",
+       "Comment on 'Death and the Gorgon'                   0.750253   \n",
+       "debating buying NVDA in 2019                        0.526975   \n",
+       "Human study on AI spear phishing campaigns          0.677458   \n",
+       "My AGI safety research—2024 review, ’25 plans       0.756925   \n",
+       "Parkinson's Law and the Ideology of Statistics      0.677458   \n",
+       "Preference Inversion                                0.633321   \n",
+       "Review: Planecrash                                  0.933950   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...  0.925483   \n",
+       "The Intelligence Curse                              0.688044   \n",
+       "The Laws of Large Numbers                           0.540932   \n",
+       "The subset parity learning problem: much more t...  0.697946   \n",
+       "What’s the short timeline plan?                     0.898161   \n",
+       "\n",
+       "                                                                               date  \\\n",
+       "title                                                                                 \n",
+       "Anthropic and Palantir Partner to Bring Claude ...        2024-07-11 00:00:00+00:00   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...        2024-12-28 00:00:00+00:00   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI                2024-12-24 00:00:00+00:00   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...        2024-12-20 00:00:00+00:00   \n",
+       "fake ai hoax paper made up by gpt-4                       2024-01-01 00:00:00+00:00   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...        2024-08-21 00:00:00+00:00   \n",
+       "How to Focus                                              2024-06-01 00:00:00+00:00   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"        2024-04-12 00:00:00+00:00   \n",
+       "By default, capital will matter more than ever ... 2024-12-30 19:47:19.838000+00:00   \n",
+       "The Plan - 2024 Update                             2024-12-31 19:29:23.013000+00:00   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...        2025-01-05 05:03:00+00:00   \n",
+       "OpenAI Email Archives from Musk v. Altman                 2024-11-01 00:00:00+00:00   \n",
+       "President Trump Announces Morgan Ortagus as Dep...        2025-01-04 00:00:00+00:00   \n",
+       "CDC Report on Missouri H5N1 Serology Testing              2025-01-05 00:00:00+00:00   \n",
+       "2024 in AI predictions                             2025-01-02 02:09:13.093000+00:00   \n",
+       "Comment on 'Death and the Gorgon'                  2025-01-01 06:00:24.498000+00:00   \n",
+       "debating buying NVDA in 2019                       2025-01-04 18:07:35.832000+00:00   \n",
+       "Human study on AI spear phishing campaigns         2025-01-03 19:03:28.406000+00:00   \n",
+       "My AGI safety research—2024 review, ’25 plans      2025-01-01 22:01:48.820000+00:00   \n",
+       "Parkinson's Law and the Ideology of Statistics     2025-01-04 22:59:57.376000+00:00   \n",
+       "Preference Inversion                               2025-01-03 23:49:06.168000+00:00   \n",
+       "Review: Planecrash                                 2025-01-01 01:31:26.864000+00:00   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh... 2025-01-01 00:42:36.538000+00:00   \n",
+       "The Intelligence Curse                             2025-01-04 18:16:58.921000+00:00   \n",
+       "The Laws of Large Numbers                          2025-01-04 18:06:02.387000+00:00   \n",
+       "The subset parity learning problem: much more t... 2025-01-04 09:59:16.158000+00:00   \n",
+       "What’s the short timeline plan?                    2025-01-05 00:10:28.708000+00:00   \n",
+       "\n",
+       "                                                    in_training  \n",
+       "title                                                            \n",
+       "Anthropic and Palantir Partner to Bring Claude ...        False  \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...        False  \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI                False  \n",
+       "Deliberative Alignment: Reasoning Enables Safer...        False  \n",
+       "fake ai hoax paper made up by gpt-4                       False  \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...        False  \n",
+       "How to Focus                                              False  \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"        False  \n",
+       "By default, capital will matter more than ever ...        False  \n",
+       "The Plan - 2024 Update                                    False  \n",
+       "Amazon-backed Anthropic debuts AI agents that c...        False  \n",
+       "OpenAI Email Archives from Musk v. Altman                 False  \n",
+       "President Trump Announces Morgan Ortagus as Dep...        False  \n",
+       "CDC Report on Missouri H5N1 Serology Testing              False  \n",
+       "2024 in AI predictions                                    False  \n",
+       "Comment on 'Death and the Gorgon'                         False  \n",
+       "debating buying NVDA in 2019                              False  \n",
+       "Human study on AI spear phishing campaigns                False  \n",
+       "My AGI safety research—2024 review, ’25 plans             False  \n",
+       "Parkinson's Law and the Ideology of Statistics            False  \n",
+       "Preference Inversion                                      False  \n",
+       "Review: Planecrash                                        False  \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...        False  \n",
+       "The Intelligence Curse                                    False  \n",
+       "The Laws of Large Numbers                                 False  \n",
+       "The subset parity learning problem: much more t...        False  \n",
+       "What’s the short timeline plan?                           False  \n",
+       "\n",
+       "[27 rows x 22 columns]"
+      ]
+     },
+     "execution_count": 18,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "# results\n",
     "df2 = pd.DataFrame(data).set_index('title')\n",
@@ -520,9 +1612,28 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 19,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "count     27.000000\n",
+       "mean     195.925926\n",
+       "std       18.097866\n",
+       "min      106.000000\n",
+       "25%      200.000000\n",
+       "50%      200.000000\n",
+       "75%      200.000000\n",
+       "max      200.000000\n",
+       "Name: summary, dtype: float64"
+      ]
+     },
+     "execution_count": 19,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "# QC summary\n",
     "df2.summary.str.len().describe()"
@@ -530,12 +1641,54 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 20,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "unsloth/Llama-3.2-1B-bnb-4bit\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "after_min            0.000692\n",
+       "improvement%_min     0.080467\n",
+       "before_min           0.114827\n",
+       "after_max            0.165648\n",
+       "improvement_min      0.179910\n",
+       "len                  0.222931\n",
+       "improvement_ppx      0.259635\n",
+       "improvement_mean     0.260877\n",
+       "improvement%_ppx     0.264841\n",
+       "before_max           0.357355\n",
+       "improvement_max      0.367002\n",
+       "before_std           0.367073\n",
+       "improvement%_max     0.369768\n",
+       "n_tokens             0.404075\n",
+       "before_mean          0.436573\n",
+       "improvement%_mean    0.442458\n",
+       "before_ppx           0.454817\n",
+       "improvement_std      0.457283\n",
+       "after_ppx            0.461407\n",
+       "after_std            0.464671\n",
+       "after_mean           0.470662\n",
+       "improvement%_std     0.482445\n",
+       "before_sum           0.574461\n",
+       "after_sum            0.580702\n",
+       "novelty              1.000000\n",
+       "Name: novelty, dtype: float64"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
    "source": [
     "for n,d in df2.groupby(\"model\"):\n",
-    "    for stat in ['mean', 'std', 'min', 'max']:\n",
+    "    for stat in ['mean', 'std', 'min', 'max', 'ppx']:\n",
     "        df2[f\"improvement%_{stat}\"] = (df2[f\"before_{stat}\"] - df2[f\"after_{stat}\"]) / df2[f\"before_{stat}\"]\n",
     "        df2[f\"improvement_{stat}\"] = (df2[f\"before_{stat}\"] - df2[f\"after_{stat}\"])\n",
     "        df2[f\"summarizable_{stat}\"] = df2[f\"improvement_{stat}\"]  > 1\n",
@@ -559,6 +1712,354 @@
    "execution_count": null,
    "metadata": {},
    "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 28,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>improvement%_mean</th>\n",
+       "      <th>novelty</th>\n",
+       "      <th>date</th>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>title</th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "      <th></th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>2024 in AI predictions</th>\n",
+       "      <td>0.026404</td>\n",
+       "      <td>0.789190</td>\n",
+       "      <td>2025-01-02 02:09:13.093000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>The Plan - 2024 Update</th>\n",
+       "      <td>0.033359</td>\n",
+       "      <td>0.785170</td>\n",
+       "      <td>2024-12-31 19:29:23.013000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>The Intelligence Curse</th>\n",
+       "      <td>0.034100</td>\n",
+       "      <td>0.688044</td>\n",
+       "      <td>2025-01-04 18:16:58.921000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>The Field of AI Alignment: A Postmortem, and What To Do About It</th>\n",
+       "      <td>0.035840</td>\n",
+       "      <td>0.925483</td>\n",
+       "      <td>2025-01-01 00:42:36.538000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Flower Crowns and Furry Mishaps by MyPalAI</th>\n",
+       "      <td>0.037607</td>\n",
+       "      <td>0.300000</td>\n",
+       "      <td>2024-12-24 00:00:00+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Preference Inversion</th>\n",
+       "      <td>0.042690</td>\n",
+       "      <td>0.633321</td>\n",
+       "      <td>2025-01-03 23:49:06.168000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>The Laws of Large Numbers</th>\n",
+       "      <td>0.047607</td>\n",
+       "      <td>0.540932</td>\n",
+       "      <td>2025-01-04 18:06:02.387000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>OpenAI Email Archives from Musk v. Altman</th>\n",
+       "      <td>0.047660</td>\n",
+       "      <td>0.700000</td>\n",
+       "      <td>2024-11-01 00:00:00+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>debating buying NVDA in 2019</th>\n",
+       "      <td>0.050691</td>\n",
+       "      <td>0.526975</td>\n",
+       "      <td>2025-01-04 18:07:35.832000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>The subset parity learning problem: much more than you wanted to know</th>\n",
+       "      <td>0.051760</td>\n",
+       "      <td>0.697946</td>\n",
+       "      <td>2025-01-04 09:59:16.158000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Hardware Hedging Against Scaling Regime Shifts (self.mlscaling)</th>\n",
+       "      <td>0.054841</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>2024-08-21 00:00:00+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Deliberative Alignment: Reasoning Enables Safer Language Models</th>\n",
+       "      <td>0.056381</td>\n",
+       "      <td>0.600000</td>\n",
+       "      <td>2024-12-20 00:00:00+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Comment on 'Death and the Gorgon'</th>\n",
+       "      <td>0.057750</td>\n",
+       "      <td>0.750253</td>\n",
+       "      <td>2025-01-01 06:00:24.498000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>By default, capital will matter more than ever after AGI</th>\n",
+       "      <td>0.060845</td>\n",
+       "      <td>0.906388</td>\n",
+       "      <td>2024-12-30 19:47:19.838000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>My AGI safety research—2024 review, ’25 plans</th>\n",
+       "      <td>0.061170</td>\n",
+       "      <td>0.756925</td>\n",
+       "      <td>2025-01-01 22:01:48.820000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Amazon-backed Anthropic debuts AI agents that can do complex tasks, racing against OpenAI, Microsoft and Google</th>\n",
+       "      <td>0.066373</td>\n",
+       "      <td>0.500000</td>\n",
+       "      <td>2025-01-05 05:03:00+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Review: Planecrash</th>\n",
+       "      <td>0.068870</td>\n",
+       "      <td>0.933950</td>\n",
+       "      <td>2025-01-01 01:31:26.864000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>CDC Report on Missouri H5N1 Serology Testing</th>\n",
+       "      <td>0.072399</td>\n",
+       "      <td>0.750000</td>\n",
+       "      <td>2025-01-05 00:00:00+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>President Trump Announces Morgan Ortagus as Deputy Special Presidential Envoy for Middle East Peace</th>\n",
+       "      <td>0.076677</td>\n",
+       "      <td>0.300000</td>\n",
+       "      <td>2025-01-04 00:00:00+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>TradingAgents: Multi-Agents LLM Financial Trading Framework</th>\n",
+       "      <td>0.076707</td>\n",
+       "      <td>0.150000</td>\n",
+       "      <td>2024-12-28 00:00:00+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>How to Focus</th>\n",
+       "      <td>0.079619</td>\n",
+       "      <td>0.500000</td>\n",
+       "      <td>2024-06-01 00:00:00+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>What’s the short timeline plan?</th>\n",
+       "      <td>0.087995</td>\n",
+       "      <td>0.898161</td>\n",
+       "      <td>2025-01-05 00:10:28.708000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Deontic Explorations In \"Paying To Talk To Slaves\"</th>\n",
+       "      <td>0.088448</td>\n",
+       "      <td>0.100000</td>\n",
+       "      <td>2024-04-12 00:00:00+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Parkinson's Law and the Ideology of Statistics</th>\n",
+       "      <td>0.089637</td>\n",
+       "      <td>0.677458</td>\n",
+       "      <td>2025-01-04 22:59:57.376000+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>fake ai hoax paper made up by gpt-4</th>\n",
+       "      <td>0.105575</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>2024-01-01 00:00:00+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Anthropic and Palantir Partner to Bring Claude AI Models to AWS for U.S. Government Intelligence and Defense Operations</th>\n",
+       "      <td>0.108802</td>\n",
+       "      <td>0.200000</td>\n",
+       "      <td>2024-07-11 00:00:00+00:00</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>Human study on AI spear phishing campaigns</th>\n",
+       "      <td>0.113187</td>\n",
+       "      <td>0.677458</td>\n",
+       "      <td>2025-01-03 19:03:28.406000+00:00</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                                                    improvement%_mean  \\\n",
+       "title                                                                   \n",
+       "2024 in AI predictions                                       0.026404   \n",
+       "The Plan - 2024 Update                                       0.033359   \n",
+       "The Intelligence Curse                                       0.034100   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...           0.035840   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI                   0.037607   \n",
+       "Preference Inversion                                         0.042690   \n",
+       "The Laws of Large Numbers                                    0.047607   \n",
+       "OpenAI Email Archives from Musk v. Altman                    0.047660   \n",
+       "debating buying NVDA in 2019                                 0.050691   \n",
+       "The subset parity learning problem: much more t...           0.051760   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...           0.054841   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...           0.056381   \n",
+       "Comment on 'Death and the Gorgon'                            0.057750   \n",
+       "By default, capital will matter more than ever ...           0.060845   \n",
+       "My AGI safety research—2024 review, ’25 plans                0.061170   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...           0.066373   \n",
+       "Review: Planecrash                                           0.068870   \n",
+       "CDC Report on Missouri H5N1 Serology Testing                 0.072399   \n",
+       "President Trump Announces Morgan Ortagus as Dep...           0.076677   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...           0.076707   \n",
+       "How to Focus                                                 0.079619   \n",
+       "What’s the short timeline plan?                              0.087995   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"           0.088448   \n",
+       "Parkinson's Law and the Ideology of Statistics               0.089637   \n",
+       "fake ai hoax paper made up by gpt-4                          0.105575   \n",
+       "Anthropic and Palantir Partner to Bring Claude ...           0.108802   \n",
+       "Human study on AI spear phishing campaigns                   0.113187   \n",
+       "\n",
+       "                                                     novelty  \\\n",
+       "title                                                          \n",
+       "2024 in AI predictions                              0.789190   \n",
+       "The Plan - 2024 Update                              0.785170   \n",
+       "The Intelligence Curse                              0.688044   \n",
+       "The Field of AI Alignment: A Postmortem, and Wh...  0.925483   \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI          0.300000   \n",
+       "Preference Inversion                                0.633321   \n",
+       "The Laws of Large Numbers                           0.540932   \n",
+       "OpenAI Email Archives from Musk v. Altman           0.700000   \n",
+       "debating buying NVDA in 2019                        0.526975   \n",
+       "The subset parity learning problem: much more t...  0.697946   \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...  1.000000   \n",
+       "Deliberative Alignment: Reasoning Enables Safer...  0.600000   \n",
+       "Comment on 'Death and the Gorgon'                   0.750253   \n",
+       "By default, capital will matter more than ever ...  0.906388   \n",
+       "My AGI safety research—2024 review, ’25 plans       0.756925   \n",
+       "Amazon-backed Anthropic debuts AI agents that c...  0.500000   \n",
+       "Review: Planecrash                                  0.933950   \n",
+       "CDC Report on Missouri H5N1 Serology Testing        0.750000   \n",
+       "President Trump Announces Morgan Ortagus as Dep...  0.300000   \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...  0.150000   \n",
+       "How to Focus                                        0.500000   \n",
+       "What’s the short timeline plan?                     0.898161   \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"  0.100000   \n",
+       "Parkinson's Law and the Ideology of Statistics      0.677458   \n",
+       "fake ai hoax paper made up by gpt-4                 0.000000   \n",
+       "Anthropic and Palantir Partner to Bring Claude ...  0.200000   \n",
+       "Human study on AI spear phishing campaigns          0.677458   \n",
+       "\n",
+       "                                                                               date  \n",
+       "title                                                                                \n",
+       "2024 in AI predictions                             2025-01-02 02:09:13.093000+00:00  \n",
+       "The Plan - 2024 Update                             2024-12-31 19:29:23.013000+00:00  \n",
+       "The Intelligence Curse                             2025-01-04 18:16:58.921000+00:00  \n",
+       "The Field of AI Alignment: A Postmortem, and Wh... 2025-01-01 00:42:36.538000+00:00  \n",
+       "Flower Crowns and Furry Mishaps by MyPalAI                2024-12-24 00:00:00+00:00  \n",
+       "Preference Inversion                               2025-01-03 23:49:06.168000+00:00  \n",
+       "The Laws of Large Numbers                          2025-01-04 18:06:02.387000+00:00  \n",
+       "OpenAI Email Archives from Musk v. Altman                 2024-11-01 00:00:00+00:00  \n",
+       "debating buying NVDA in 2019                       2025-01-04 18:07:35.832000+00:00  \n",
+       "The subset parity learning problem: much more t... 2025-01-04 09:59:16.158000+00:00  \n",
+       "Hardware Hedging Against Scaling Regime Shifts ...        2024-08-21 00:00:00+00:00  \n",
+       "Deliberative Alignment: Reasoning Enables Safer...        2024-12-20 00:00:00+00:00  \n",
+       "Comment on 'Death and the Gorgon'                  2025-01-01 06:00:24.498000+00:00  \n",
+       "By default, capital will matter more than ever ... 2024-12-30 19:47:19.838000+00:00  \n",
+       "My AGI safety research—2024 review, ’25 plans      2025-01-01 22:01:48.820000+00:00  \n",
+       "Amazon-backed Anthropic debuts AI agents that c...        2025-01-05 05:03:00+00:00  \n",
+       "Review: Planecrash                                 2025-01-01 01:31:26.864000+00:00  \n",
+       "CDC Report on Missouri H5N1 Serology Testing              2025-01-05 00:00:00+00:00  \n",
+       "President Trump Announces Morgan Ortagus as Dep...        2025-01-04 00:00:00+00:00  \n",
+       "TradingAgents: Multi-Agents LLM Financial Tradi...        2024-12-28 00:00:00+00:00  \n",
+       "How to Focus                                              2024-06-01 00:00:00+00:00  \n",
+       "What’s the short timeline plan?                    2025-01-05 00:10:28.708000+00:00  \n",
+       "Deontic Explorations In \"Paying To Talk To Slaves\"        2024-04-12 00:00:00+00:00  \n",
+       "Parkinson's Law and the Ideology of Statistics     2025-01-04 22:59:57.376000+00:00  \n",
+       "fake ai hoax paper made up by gpt-4                       2024-01-01 00:00:00+00:00  \n",
+       "Anthropic and Palantir Partner to Bring Claude ...        2024-07-11 00:00:00+00:00  \n",
+       "Human study on AI spear phishing campaigns         2025-01-03 19:03:28.406000+00:00  "
+      ]
+     },
+     "execution_count": 28,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "df2[[ 'improvement%_mean', 'novelty', 'date']].sort_values(\"improvement%_mean\", ascending=True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "improvement%_std    -0.482445\n",
+       "improvement_std     -0.457283\n",
+       "improvement%_mean   -0.442458\n",
+       "improvement%_max    -0.369768\n",
+       "improvement_max     -0.367002\n",
+       "before_max          -0.357355\n",
+       "improvement%_ppx    -0.264841\n",
+       "improvement_mean    -0.260877\n",
+       "after_max           -0.165648\n",
+       "improvement%_min    -0.080467\n",
+       "after_min           -0.000692\n",
+       "before_min           0.114827\n",
+       "improvement_min      0.179910\n",
+       "len                  0.222931\n",
+       "improvement_ppx      0.259635\n",
+       "before_std           0.367073\n",
+       "n_tokens             0.404075\n",
+       "before_mean          0.436573\n",
+       "before_ppx           0.454817\n",
+       "after_ppx            0.461407\n",
+       "after_std            0.464671\n",
+       "after_mean           0.470662\n",
+       "before_sum           0.574461\n",
+       "after_sum            0.580702\n",
+       "novelty              1.000000\n",
+       "Name: novelty, dtype: float64"
+      ]
+     },
+     "execution_count": 21,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "df2.select_dtypes(include=np.number).corr()['novelty'].sort_values()"
    ]
diff --git a/nbs/03_get_lesswrong_data.ipynb b/nbs/03_get_lesswrong_data.ipynb
index 0cf2600..729771b 100644
--- a/nbs/03_get_lesswrong_data.ipynb
+++ b/nbs/03_get_lesswrong_data.ipynb
@@ -470,7 +470,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 144,
+   "execution_count": 147,
    "metadata": {},
    "outputs": [
     {
@@ -486,7 +486,27 @@
       "../samples/2025_lw_the-intelligence-curse.md 1.2061121463775635\n",
       "../samples/2025_lw_the-intelligence-curse.md 1.2061121463775635\n",
       "../samples/2025_lw_human-study-on-ai-spear-phishing-campaigns.md 0.9995136260986328\n",
-      "../samples/2025_lw_human-study-on-ai-spear-phishing-campaigns.md 0.9995136260986328\n"
+      "../samples/2025_lw_human-study-on-ai-spear-phishing-campaigns.md 0.9995136260986328\n",
+      "../samples/2025_lw_the-subset-parity-learning-problem-much-more-than-you-wanted.md 0.9548193216323853\n",
+      "../samples/2025_lw_the-subset-parity-learning-problem-much-more-than-you-wanted.md 0.9548193216323853\n",
+      "../samples/2025_lw_2024-in-ai-predictions.md 0.8065339922904968\n",
+      "../samples/2025_lw_2024-in-ai-predictions.md 0.8065339922904968\n",
+      "../samples/2025_lw_debating-buying-nvda-in-2019.md 0.7926478385925293\n",
+      "../samples/2025_lw_debating-buying-nvda-in-2019.md 0.7926478385925293\n",
+      "../samples/2025_lw_review-planecrash.md 0.689734160900116\n",
+      "../samples/2025_lw_review-planecrash.md 0.689734160900116\n",
+      "../samples/2024_lw_by-default-capital-will-matter-more-than-ever-after-agi.md 0.6629015207290649\n",
+      "../samples/2024_lw_by-default-capital-will-matter-more-than-ever-after-agi.md 0.6629015207290649\n",
+      "../samples/2025_lw_the-field-of-ai-alignment-a-postmortem-and-what-to-do-about.md 0.5714353919029236\n",
+      "../samples/2025_lw_the-field-of-ai-alignment-a-postmortem-and-what-to-do-about.md 0.5714353919029236\n",
+      "../samples/2024_lw_the-plan-2024-update.md 0.542655885219574\n",
+      "../samples/2024_lw_the-plan-2024-update.md 0.542655885219574\n",
+      "../samples/2025_lw_comment-on-death-and-the-gorgon.md 0.5308915376663208\n",
+      "../samples/2025_lw_comment-on-death-and-the-gorgon.md 0.5308915376663208\n",
+      "../samples/2025_lw_my-agi-safety-research-2024-review-25-plans.md 0.49594494700431824\n",
+      "../samples/2025_lw_my-agi-safety-research-2024-review-25-plans.md 0.49594494700431824\n",
+      "../samples/2025_lw_preference-inversion.md 0.48199906945228577\n",
+      "../samples/2025_lw_preference-inversion.md 0.48199906945228577\n"
      ]
     }
    ],
@@ -495,7 +515,7 @@
     "    md = markdownify(row[\"htmlBody\"]).strip()\n",
     "\n",
     "    return f\"\"\"---\n",
-    "title: {row['title']}\n",
+    "title: \"{row['title'].replace('\"', \"'\")}\"\n",
     "date: {row['modifiedAt']}\n",
     "url: {row['pageUrl']}\n",
     "novelty: {row['novelty']}\n",
@@ -507,7 +527,7 @@
     "\"\"\"\n",
     "\n",
     "\n",
-    "for i in range(5):\n",
+    "for i in range(15):\n",
     "    for ii in [i, -i-1]:\n",
     "        row = df.iloc[i]\n",
     "        s = to_markdown(row)\n",