wassname/peft
1
0
Fork 0
mirror of https://github.com/wassname/peft.git synced 2026-09-11 12:30:16 +08:00
Code Issues Packages Projects Releases Wiki Activity

add examples and update README

Browse Source
This commit is contained in:
Sourab Mangrulkar
2022-12-29 17:34:39 +05:30
parent 5de1751843
commit f16750afcc
10 changed files with 5218 additions and 89 deletions
Download Patch File Download Diff File Expand all files Collapse all files
examples/conditional_generation/pet_lora_seq2seq.ipynb
+410
View File
@@ -0,0 +1,410 @@
{
"cells": [
{
"cell_type": "code",
"execution_count": 17,
"id": "5f93b7d1",
"metadata": {},
"outputs": [],
"source": [
"from transformers import AutoModelForSeq2SeqLM\n",
"from pet import get_pet_config,get_pet_model, get_pet_model_state_dict, LoRAConfig, TaskType\n",
"import torch\n",
"from datasets import load_dataset\n",
"import os\n",
"os.environ[\"TOKENIZERS_PARALLELISM\"] = \"false\"\n",
"from transformers import AutoTokenizer\n",
"from torch.utils.data import DataLoader\n",
"from transformers import default_data_collator,get_linear_schedule_with_warmup\n",
"from tqdm import tqdm\n",
"from datasets import load_dataset\n",
"\n",
"device = \"cuda\"\n",
"model_name_or_path = \"bigscience/mt0-large\"\n",
"tokenizer_name_or_path = \"bigscience/mt0-large\"\n",
"\n",
"checkpoint_name = \"financial_sentiment_analysis_lora_v1.pt\"\n",
"text_column = \"sentence\"\n",
"label_column = \"text_label\"\n",
"max_length=128\n",
"lr = 1e-3\n",
"num_epochs = 3\n",
"batch_size=8\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "8d0850ac",
"metadata": {},
"outputs": [],
"source": [
"# creating model\n",
"pet_config = LoRAConfig(\n",
" task_type=TaskType.SEQ_2_SEQ_LM, inference_mode=False, r=8, lora_alpha=32, lora_dropout=0.1\n",
")\n",
"\n",
"model = AutoModelForSeq2SeqLM.from_pretrained(model_name_or_path)\n",
"model = get_pet_model(model, pet_config)\n",
"model.print_trainable_parameters()\n",
"model"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "4ee2babf",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/sourab/miniconda3/envs/ml/lib/python3.10/site-packages/huggingface_hub/utils/_deprecation.py:97: FutureWarning: Deprecated argument(s) used in 'dataset_info': token. Will not be supported from version '0.12'.\n",
" warnings.warn(message, FutureWarning)\n",
"Found cached dataset financial_phrasebank (/home/sourab/.cache/huggingface/datasets/financial_phrasebank/sentences_allagree/1.0.0/550bde12e6c30e2674da973a55f57edde5181d53f5a5a34c1531c53f93b7e141)\n"
]
},
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "6de075f8208349108291ac5ab7f5c980",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
" 0%| | 0/1 [00:00<?, ?it/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "4b0e67b6d93f43e4b0f6a2f8978e4b0c",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
" 0%| | 0/3 [00:00<?, ?ba/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "a9551029c9884529bda7421a99170b51",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
" 0%| | 0/1 [00:00<?, ?ba/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/plain": [
"{'sentence': 'The order was valued at USD12 .2 m.',\n",
" 'label': 1,\n",
" 'text_label': 'neutral'}"
]
},
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# loading dataset\n",
"dataset = load_dataset(\"financial_phrasebank\", 'sentences_allagree')\n",
"dataset = dataset[\"train\"].train_test_split(test_size=0.1)\n",
"dataset[\"validation\"] = dataset[\"test\"]\n",
"del(dataset[\"test\"])\n",
"\n",
"classes = dataset[\"train\"].features[\"label\"].names\n",
"dataset = dataset.map(\n",
" lambda x: {\"text_label\": [classes[label] for label in x[\"label\"]]},\n",
" batched=True,\n",
" num_proc=1,\n",
" \n",
")\n",
"\n",
"dataset[\"train\"][0]"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "adf9608c",
"metadata": {},
"outputs": [
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "4421971232434db1b6141e91fda2f6d7",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Running tokenizer on dataset: 0%| | 0/3 [00:00<?, ?ba/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "9b2ef793d93443949f4a5d5874d4bc05",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Running tokenizer on dataset: 0%| | 0/1 [00:00<?, ?ba/s]"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# data preprocessing\n",
"tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)\n",
"def preprocess_function(examples):\n",
" inputs = examples[text_column]\n",
" targets = examples[label_column]\n",
" model_inputs = tokenizer(inputs, max_length=max_length, padding=\"max_length\", truncation=True, return_tensors=\"pt\")\n",
" labels = tokenizer(targets, max_length=3, padding=\"max_length\", truncation=True, return_tensors=\"pt\")\n",
" labels = labels[\"input_ids\"]\n",
" labels[labels==tokenizer.pad_token_id] = -100\n",
" model_inputs[\"labels\"] = labels\n",
" return model_inputs\n",
"\n",
"processed_datasets = dataset.map(\n",
" preprocess_function,\n",
" batched=True,\n",
" num_proc=1,\n",
" remove_columns=dataset[\"train\"].column_names,\n",
" load_from_cache_file=False,\n",
" desc=\"Running tokenizer on dataset\",\n",
" )\n",
"\n",
"train_dataset = processed_datasets[\"train\"]\n",
"eval_dataset = processed_datasets[\"validation\"]\n",
"\n",
"train_dataloader = DataLoader(\n",
" train_dataset, shuffle=True, collate_fn=default_data_collator, batch_size=batch_size, pin_memory=True\n",
" )\n",
"eval_dataloader = DataLoader(eval_dataset, collate_fn=default_data_collator, batch_size=batch_size, pin_memory=True)\n",
"\n",
"\n",
"\n",
" "
]
},
{
"cell_type": "code",
"execution_count": 5,
"id": "f733a3c6",
"metadata": {},
"outputs": [],
"source": [
"# optimizer and lr scheduler\n",
"optimizer = torch.optim.AdamW(model.parameters(), lr=lr)\n",
"lr_scheduler = get_linear_schedule_with_warmup(\n",
" optimizer=optimizer,\n",
" num_warmup_steps=0,\n",
" num_training_steps=(len(train_dataloader) * num_epochs),\n",
")\n"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "6b3a4090",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|█████████████████████████████████████████████████████████████| 255/255 [00:53<00:00, 4.80it/s]\n",
"100%|███████████████████████████████████████████████████████████████| 29/29 [00:02<00:00, 14.16it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"epoch=0: train_ppl=tensor(13.6966, device='cuda:0') train_epoch_loss=tensor(2.6171, device='cuda:0') eval_ppl=tensor(1.0046, device='cuda:0') eval_epoch_loss=tensor(0.0046, device='cuda:0')\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|█████████████████████████████████████████████████████████████| 255/255 [00:52<00:00, 4.88it/s]\n",
"100%|███████████████████████████████████████████████████████████████| 29/29 [00:02<00:00, 14.20it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"epoch=1: train_ppl=tensor(1.5893, device='cuda:0') train_epoch_loss=tensor(0.4633, device='cuda:0') eval_ppl=tensor(1.0020, device='cuda:0') eval_epoch_loss=tensor(0.0020, device='cuda:0')\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|█████████████████████████████████████████████████████████████| 255/255 [00:52<00:00, 4.87it/s]\n",
"100%|███████████████████████████████████████████████████████████████| 29/29 [00:02<00:00, 14.18it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"epoch=2: train_ppl=tensor(1.3210, device='cuda:0') train_epoch_loss=tensor(0.2784, device='cuda:0') eval_ppl=tensor(1.0026, device='cuda:0') eval_epoch_loss=tensor(0.0026, device='cuda:0')\n"
]
}
],
"source": [
"# training and evaluation\n",
"model = model.to(device)\n",
"\n",
"for epoch in range(num_epochs):\n",
" model.train()\n",
" total_loss = 0\n",
" for step, batch in enumerate(tqdm(train_dataloader)):\n",
" batch = {k: v.to(device) for k, v in batch.items()}\n",
" outputs = model(**batch)\n",
" loss = outputs.loss\n",
" total_loss += loss.detach().float()\n",
" loss.backward()\n",
" optimizer.step()\n",
" lr_scheduler.step()\n",
" optimizer.zero_grad()\n",
"\n",
" model.eval()\n",
" eval_loss = 0\n",
" eval_preds = []\n",
" for step, batch in enumerate(tqdm(eval_dataloader)):\n",
" batch = {k: v.to(device) for k, v in batch.items()}\n",
" with torch.no_grad():\n",
" outputs = model(**batch)\n",
" loss = outputs.loss\n",
" eval_loss += loss.detach().float()\n",
" eval_preds.extend(tokenizer.batch_decode(torch.argmax(outputs.logits, -1).detach().cpu().numpy(), skip_special_tokens=True))\n",
"\n",
" eval_epoch_loss = eval_loss/len(train_dataloader)\n",
" eval_ppl = torch.exp(eval_epoch_loss)\n",
" train_epoch_loss = total_loss/len(eval_dataloader)\n",
" train_ppl = torch.exp(train_epoch_loss)\n",
" print(f\"{epoch=}: {train_ppl=} {train_epoch_loss=} {eval_ppl=} {eval_epoch_loss=}\")\n"
]
},
{
"cell_type": "code",
"execution_count": 20,
"id": "6cafa67b",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"accuracy=98.23788546255507 % on the evaluation dataset\n",
"eval_preds[:10]=['neutral', 'neutral', 'positive', 'positive', 'neutral', 'neutral', 'neutral', 'neutral', 'neutral', 'neutral']\n",
"dataset['validation']['text_label'][:10]=['neutral', 'neutral', 'positive', 'positive', 'neutral', 'neutral', 'neutral', 'neutral', 'neutral', 'neutral']\n"
]
}
],
"source": [
"# print accuracy\n",
"correct =0\n",
"total = 0\n",
"for pred,true in zip(eval_preds, dataset[\"validation\"][\"text_label\"]):\n",
" if pred.strip()==true.strip():\n",
" correct+=1\n",
" total+=1 \n",
"accuracy = correct/total*100\n",
"print(f\"{accuracy=} % on the evaluation dataset\")\n",
"print(f\"{eval_preds[:10]=}\")\n",
"print(f\"{dataset['validation']['text_label'][:10]=}\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "a8de6005",
"metadata": {},
"outputs": [],
"source": [
"# saving model\n",
"state_dict = get_pet_model_state_dict(model)\n",
"torch.save(state_dict, checkpoint_name)\n",
"print(state_dict)"
]
},
{
"cell_type": "code",
"execution_count": 18,
"id": "bd20cd4c",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"19M\tfinancial_sentiment_analysis_lora_v1.pt\r\n"
]
}
],
"source": [
"!du -h $checkpoint_name"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "76c2fc29",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3.10.5 64-bit",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.10.5 (v3.10.5:f377153967, Jun 6 2022, 12:36:10) [Clang 13.0.0 (clang-1300.0.29.30)]"
},
"vscode": {
"interpreter": {
"hash": "aee8b7b246df8f9039afb4144a1f6fd8d2ca17a180786b69acc140d282b71a49"
}
}
},
"nbformat": 4,
"nbformat_minor": 5
}
examples/conditional_generation/pet_lora_seq2seq_accelerate_ds_zero3_offload.py
+297
View File
@@ -0,0 +1,297 @@
import gc
import os
import sys
import threading
import numpy as np
import torch
from accelerate import Accelerator
from torch.utils.data import DataLoader
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer, get_linear_schedule_with_warmup, set_seed
import psutil
from datasets import load_dataset
from pet import LoRAConfig, TaskType, get_pet_model, get_pet_model_state_dict
from tqdm import tqdm
def levenshtein_distance(str1, str2):
# TC: O(N^2)
# SC: O(N^2)
if str1 == str2:
return 0
num_rows = len(str1) + 1
num_cols = len(str2) + 1
dp_matrix = np.empty((num_rows, num_cols))
dp_matrix[0, :] = range(num_cols)
dp_matrix[:, 0] = range(num_rows)
for i in range(1, num_rows):
for j in range(1, num_cols):
if str1[i - 1] == str2[j - 1]:
dp_matrix[i, j] = dp_matrix[i - 1, j - 1]
else:
dp_matrix[i, j] = min(dp_matrix[i - 1, j - 1], dp_matrix[i - 1, j], dp_matrix[i, j - 1]) + 1
return dp_matrix[num_rows - 1, num_cols - 1]
def get_closest_label(eval_pred, classes):
min_id = sys.maxsize
min_edit_distance = sys.maxsize
for i, class_label in enumerate(classes):
edit_distance = levenshtein_distance(eval_pred.strip(), class_label)
if edit_distance < min_edit_distance:
min_id = i
min_edit_distance = edit_distance
return classes[min_id]
# Converting Bytes to Megabytes
def b2mb(x):
return int(x / 2**20)
# This context manager is used to track the peak memory usage of the process
class TorchTracemalloc:
def __enter__(self):
gc.collect()
torch.cuda.empty_cache()
torch.cuda.reset_max_memory_allocated() # reset the peak gauge to zero
self.begin = torch.cuda.memory_allocated()
self.process = psutil.Process()
self.cpu_begin = self.cpu_mem_used()
self.peak_monitoring = True
peak_monitor_thread = threading.Thread(target=self.peak_monitor_func)
peak_monitor_thread.daemon = True
peak_monitor_thread.start()
return self
def cpu_mem_used(self):
"""get resident set size memory for the current process"""
return self.process.memory_info().rss
def peak_monitor_func(self):
self.cpu_peak = -1
while True:
self.cpu_peak = max(self.cpu_mem_used(), self.cpu_peak)
# can't sleep or will not catch the peak right (this comment is here on purpose)
# time.sleep(0.001) # 1msec
if not self.peak_monitoring:
break
def __exit__(self, *exc):
self.peak_monitoring = False
gc.collect()
torch.cuda.empty_cache()
self.end = torch.cuda.memory_allocated()
self.peak = torch.cuda.max_memory_allocated()
self.used = b2mb(self.end - self.begin)
self.peaked = b2mb(self.peak - self.begin)
self.cpu_end = self.cpu_mem_used()
self.cpu_used = b2mb(self.cpu_end - self.cpu_begin)
self.cpu_peaked = b2mb(self.cpu_peak - self.cpu_begin)
# print(f"delta used/peak {self.used:4d}/{self.peaked:4d}")
def main():
accelerator = Accelerator()
model_name_or_path = "bigscience/T0_3B"
dataset_name = "twitter_complaints"
pet_config = LoRAConfig(
task_type=TaskType.SEQ_2_SEQ_LM, inference_mode=False, r=8, lora_alpha=32, lora_dropout=0.1
)
checkpoint_name = f"{dataset_name}_{pet_config.pet_type}_{pet_config.task_type}_v1.pt".replace("/", "_")
text_column = "Tweet text"
label_column = "text_label"
lr = 3e-3
num_epochs = 5
batch_size = 8
seed = 42
set_seed(seed)
dataset = load_dataset("ought/raft", dataset_name)
classes = [k.replace("_", " ") for k in dataset["train"].features["Label"].names]
dataset = dataset.map(
lambda x: {"text_label": [classes[label] for label in x["Label"]]},
batched=True,
num_proc=1,
)
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)
target_max_length = max([len(tokenizer(class_label)["input_ids"]) for class_label in classes])
def preprocess_function(examples):
inputs = examples[text_column]
targets = examples[label_column]
model_inputs = tokenizer(inputs, truncation=True)
labels = tokenizer(
targets, max_length=target_max_length, padding="max_length", truncation=True, return_tensors="pt"
)
labels = labels["input_ids"]
labels[labels == tokenizer.pad_token_id] = -100
model_inputs["labels"] = labels
return model_inputs
with accelerator.main_process_first():
processed_datasets = dataset.map(
preprocess_function,
batched=True,
num_proc=1,
remove_columns=dataset["train"].column_names,
load_from_cache_file=True,
desc="Running tokenizer on dataset",
)
accelerator.wait_for_everyone()
train_dataset = processed_datasets["train"]
eval_dataset = processed_datasets["train"]
test_dataset = processed_datasets["test"]
def collate_fn(examples):
return tokenizer.pad(examples, padding="longest", return_tensors="pt")
train_dataloader = DataLoader(
train_dataset, shuffle=True, collate_fn=collate_fn, batch_size=batch_size, pin_memory=True
)
eval_dataloader = DataLoader(eval_dataset, collate_fn=collate_fn, batch_size=batch_size, pin_memory=True)
test_dataloader = DataLoader(test_dataset, collate_fn=collate_fn, batch_size=batch_size, pin_memory=True)
# creating model
model = AutoModelForSeq2SeqLM.from_pretrained(model_name_or_path)
model = get_pet_model(model, pet_config)
model.print_trainable_parameters()
# optimizer
optimizer = torch.optim.AdamW(model.parameters(), lr=lr)
# lr scheduler
lr_scheduler = get_linear_schedule_with_warmup(
optimizer=optimizer,
num_warmup_steps=0,
num_training_steps=(len(train_dataloader) * num_epochs),
)
model, train_dataloader, eval_dataloader, test_dataloader, optimizer, lr_scheduler = accelerator.prepare(
model, train_dataloader, eval_dataloader, test_dataloader, optimizer, lr_scheduler
)
accelerator.print(model)
is_ds_zero_3 = False
if getattr(accelerator.state, "deepspeed_plugin", None):
is_ds_zero_3 = accelerator.state.deepspeed_plugin.zero_stage == 3
for epoch in range(num_epochs):
with TorchTracemalloc() as tracemalloc:
model.train()
total_loss = 0
for step, batch in enumerate(tqdm(train_dataloader)):
outputs = model(**batch)
loss = outputs.loss
total_loss += loss.detach().float()
accelerator.backward(loss)
optimizer.step()
lr_scheduler.step()
optimizer.zero_grad()
# Printing the GPU memory usage details such as allocated memory, peak memory, and total memory usage
accelerator.print("GPU Memory before entering the train : {}".format(b2mb(tracemalloc.begin)))
accelerator.print("GPU Memory consumed at the end of the train (end-begin): {}".format(tracemalloc.used))
accelerator.print("GPU Peak Memory consumed during the train (max-begin): {}".format(tracemalloc.peaked))
accelerator.print(
"GPU Total Peak Memory consumed during the train (max): {}".format(
tracemalloc.peaked + b2mb(tracemalloc.begin)
)
)
accelerator.print("CPU Memory before entering the train : {}".format(b2mb(tracemalloc.cpu_begin)))
accelerator.print("CPU Memory consumed at the end of the train (end-begin): {}".format(tracemalloc.cpu_used))
accelerator.print("CPU Peak Memory consumed during the train (max-begin): {}".format(tracemalloc.cpu_peaked))
accelerator.print(
"CPU Total Peak Memory consumed during the train (max): {}".format(
tracemalloc.cpu_peaked + b2mb(tracemalloc.cpu_begin)
)
)
train_epoch_loss = total_loss / len(eval_dataloader)
train_ppl = torch.exp(train_epoch_loss)
accelerator.print(f"{epoch=}: {train_ppl=} {train_epoch_loss=}")
model.eval()
eval_preds = []
with TorchTracemalloc() as tracemalloc:
for _, batch in enumerate(tqdm(eval_dataloader)):
batch = {k: v for k, v in batch.items() if k != "labels"}
with torch.no_grad():
outputs = accelerator.unwrap_model(model).generate(
**batch, synced_gpus=is_ds_zero_3
) # synced_gpus=True for DS-stage 3
preds = outputs.detach().cpu().numpy()
eval_preds.extend(tokenizer.batch_decode(preds, skip_special_tokens=True))
# Printing the GPU memory usage details such as allocated memory, peak memory, and total memory usage
accelerator.print("GPU Memory before entering the eval : {}".format(b2mb(tracemalloc.begin)))
accelerator.print("GPU Memory consumed at the end of the eval (end-begin): {}".format(tracemalloc.used))
accelerator.print("GPU Peak Memory consumed during the eval (max-begin): {}".format(tracemalloc.peaked))
accelerator.print(
"GPU Total Peak Memory consumed during the eval (max): {}".format(
tracemalloc.peaked + b2mb(tracemalloc.begin)
)
)
accelerator.print("CPU Memory before entering the eval : {}".format(b2mb(tracemalloc.cpu_begin)))
accelerator.print("CPU Memory consumed at the end of the eval (end-begin): {}".format(tracemalloc.cpu_used))
accelerator.print("CPU Peak Memory consumed during the eval (max-begin): {}".format(tracemalloc.cpu_peaked))
accelerator.print(
"CPU Total Peak Memory consumed during the eval (max): {}".format(
tracemalloc.cpu_peaked + b2mb(tracemalloc.cpu_begin)
)
)
correct = 0
total = 0
for pred, true in zip(eval_preds, dataset["train"][label_column]):
if pred.strip() == true.strip():
correct += 1
total += 1
accuracy = correct / total * 100
accelerator.print(f"{accuracy=}")
accelerator.print(f"{eval_preds[:10]=}")
accelerator.print(f"{dataset['train'][label_column][:10]=}")
model.eval()
test_preds = []
for _, batch in enumerate(tqdm(test_dataloader)):
batch = {k: v for k, v in batch.items() if k != "labels"}
with torch.no_grad():
outputs = accelerator.unwrap_model(model).generate(
**batch, synced_gpus=is_ds_zero_3
) # synced_gpus=True for DS-stage 3
test_preds.extend(tokenizer.batch_decode(outputs.detach().cpu().numpy(), skip_special_tokens=True))
test_preds_cleaned = []
for _, pred in enumerate(test_preds):
test_preds_cleaned.append(get_closest_label(pred, classes))
test_df = dataset["test"].to_pandas()
test_df[label_column] = test_preds_cleaned
test_df["text_labels_orig"] = test_preds
accelerator.print(test_df[[text_column, label_column]].sample(20))
pred_df = test_df[["ID", label_column]]
pred_df.columns = ["ID", "Label"]
os.makedirs(f"data/{dataset_name}", exist_ok=True)
pred_df.to_csv(f"data/{dataset_name}/predictions.csv", index=False)
accelerator.wait_for_everyone()
accelerator.save(get_pet_model_state_dict(model, state_dict=accelerator.get_state_dict(model)), checkpoint_name)
accelerator.wait_for_everyone()
if __name__ == "__main__":
main()
examples/conditional_generation/pet_lora_seq2seq_accelerate_fsdp.py
+136
View File
@@ -0,0 +1,136 @@
import os
import torch
from accelerate import Accelerator
from torch.utils.data import DataLoader
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer, default_data_collator, get_linear_schedule_with_warmup
from datasets import load_dataset
from pet import LoRAConfig, TaskType, get_pet_model, get_pet_model_state_dict
from pet.utils.other import fsdp_auto_wrap_policy
from tqdm import tqdm
def main():
accelerator = Accelerator()
model_name_or_path = "t5-base"
batch_size = 8
text_column = "sentence"
label_column = "label"
max_length = 64
lr = 1e-3
num_epochs = 1
base_path = "temp/data/FinancialPhraseBank-v1.0"
pet_config = LoRAConfig(
task_type=TaskType.SEQ_2_SEQ_LM, inference_mode=False, r=8, lora_alpha=32, lora_dropout=0.1
)
checkpoint_name = "financial_sentiment_analysis_lora_fsdp_v1.pt"
model = AutoModelForSeq2SeqLM.from_pretrained(model_name_or_path)
model = get_pet_model(model, pet_config)
accelerator.print(model.print_trainable_parameters())
dataset = load_dataset(
"json",
data_files={
"train": os.path.join(base_path, "financial_phrase_bank_train.jsonl"),
"validation": os.path.join(base_path, "financial_phrase_bank_val.jsonl"),
},
)
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)
def preprocess_function(examples):
inputs = examples[text_column]
targets = examples[label_column]
model_inputs = tokenizer(
inputs, max_length=max_length, padding="max_length", truncation=True, return_tensors="pt"
)
labels = tokenizer(targets, max_length=2, padding="max_length", truncation=True, return_tensors="pt")
labels = labels["input_ids"]
labels[labels == tokenizer.pad_token_id] = -100
model_inputs["labels"] = labels
return model_inputs
with accelerator.main_process_first():
processed_datasets = dataset.map(
preprocess_function,
batched=True,
num_proc=1,
remove_columns=dataset["train"].column_names,
load_from_cache_file=False,
desc="Running tokenizer on dataset",
)
train_dataset = processed_datasets["train"]
eval_dataset = processed_datasets["validation"]
train_dataloader = DataLoader(
train_dataset, shuffle=True, collate_fn=default_data_collator, batch_size=batch_size, pin_memory=True
)
eval_dataloader = DataLoader(
eval_dataset, collate_fn=default_data_collator, batch_size=batch_size, pin_memory=True
)
optimizer = torch.optim.AdamW(model.parameters(), lr=lr)
lr_scheduler = get_linear_schedule_with_warmup(
optimizer=optimizer,
num_warmup_steps=0,
num_training_steps=(len(train_dataloader) * num_epochs),
)
if getattr(accelerator.state, "fsdp_plugin", None) is not None:
accelerator.state.fsdp_plugin.auto_wrap_policy = fsdp_auto_wrap_policy(model)
model, train_dataloader, eval_dataloader, optimizer, lr_scheduler = accelerator.prepare(
model, train_dataloader, eval_dataloader, optimizer, lr_scheduler
)
accelerator.print(model)
for epoch in range(num_epochs):
model.train()
total_loss = 0
for step, batch in enumerate(tqdm(train_dataloader)):
outputs = model(**batch)
loss = outputs.loss
total_loss += loss.detach().float()
loss.backward()
optimizer.step()
lr_scheduler.step()
optimizer.zero_grad()
model.eval()
eval_loss = 0
eval_preds = []
for step, batch in enumerate(tqdm(eval_dataloader)):
with torch.no_grad():
outputs = model(**batch)
loss = outputs.loss
eval_loss += loss.detach().float()
preds = accelerator.gather_for_metrics(torch.argmax(outputs.logits, -1)).detach().cpu().numpy()
eval_preds.extend(tokenizer.batch_decode(preds, skip_special_tokens=True))
eval_epoch_loss = eval_loss / len(train_dataloader)
eval_ppl = torch.exp(eval_epoch_loss)
train_epoch_loss = total_loss / len(eval_dataloader)
train_ppl = torch.exp(train_epoch_loss)
accelerator.print(f"{epoch=}: {train_ppl=} {train_epoch_loss=} {eval_ppl=} {eval_epoch_loss=}")
correct = 0
total = 0
for pred, true in zip(eval_preds, dataset["validation"][label_column]):
if pred.strip() == true.strip():
correct += 1
total += 1
accuracy = correct / total * 100
accelerator.print(f"{accuracy=}")
accelerator.print(f"{eval_preds[:10]=}")
accelerator.print(f"{dataset['validation'][label_column][:10]=}")
accelerator.wait_for_everyone()
accelerator.save(
get_pet_model_state_dict(model, state_dict=accelerator.get_state_dict(model)), checkpoint_name
)
accelerator.wait_for_everyone()
if __name__ == "__main__":
main()
examples/conditional_generation/pet_prefix_tuning_seq2seq.ipynb
+502
View File
@@ -0,0 +1,502 @@
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"id": "5f93b7d1",
"metadata": {},
"outputs": [],
"source": [
"from transformers import AutoModelForSeq2SeqLM\n",
"from pet import get_pet_config,get_pet_model, get_pet_model_state_dict, PrefixTuningConfig, TaskType\n",
"import torch\n",
"from datasets import load_dataset\n",
"import os\n",
"os.environ[\"TOKENIZERS_PARALLELISM\"] = \"false\"\n",
"os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"3\"\n",
"from transformers import AutoTokenizer\n",
"from torch.utils.data import DataLoader\n",
"from transformers import default_data_collator,get_linear_schedule_with_warmup\n",
"from tqdm import tqdm\n",
"from datasets import load_dataset\n",
"\n",
"device = \"cuda\"\n",
"model_name_or_path = \"t5-large\"\n",
"tokenizer_name_or_path = \"t5-large\"\n",
"\n",
"checkpoint_name = \"financial_sentiment_analysis_prefix_tuning_v1.pt\"\n",
"text_column = \"sentence\"\n",
"label_column = \"text_label\"\n",
"max_length=128\n",
"lr = 1e-2\n",
"num_epochs = 5\n",
"batch_size=8\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "8d0850ac",
"metadata": {},
"outputs": [],
"source": [
"# creating model\n",
"pet_config = PrefixTuningConfig(\n",
" task_type=TaskType.SEQ_2_SEQ_LM, inference_mode=False, num_virtual_tokens=20\n",
")\n",
"\n",
"model = AutoModelForSeq2SeqLM.from_pretrained(model_name_or_path)\n",
"model = get_pet_model(model, pet_config)\n",
"model.print_trainable_parameters()\n",
"model"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "4ee2babf",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/sourab/miniconda3/envs/ml/lib/python3.10/site-packages/huggingface_hub/utils/_deprecation.py:97: FutureWarning: Deprecated argument(s) used in 'dataset_info': token. Will not be supported from version '0.12'.\n",
" warnings.warn(message, FutureWarning)\n",
"Found cached dataset financial_phrasebank (/home/sourab/.cache/huggingface/datasets/financial_phrasebank/sentences_allagree/1.0.0/550bde12e6c30e2674da973a55f57edde5181d53f5a5a34c1531c53f93b7e141)\n"
]
},
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "e3f8b8faca0a4112b2c3499faee9544b",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
" 0%| | 0/1 [00:00<?, ?it/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "935c8aebde284a5784348588e0bb013a",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
" 0%| | 0/3 [00:00<?, ?ba/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "e3487cd55f6847588492bf7fa51348ca",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
" 0%| | 0/1 [00:00<?, ?ba/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/plain": [
"{'sentence': 'ADPnews - Feb 5 , 2010 - Finnish real estate investor Sponda Oyj HEL : SDA1V said today that it slipped to a net loss of EUR 81.5 million USD 11.8 m in 2009 from a profit of EUR 29.3 million in 2008 .',\n",
" 'label': 0,\n",
" 'text_label': 'negative'}"
]
},
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# loading dataset\n",
"dataset = load_dataset(\"financial_phrasebank\", 'sentences_allagree')\n",
"dataset = dataset[\"train\"].train_test_split(test_size=0.1)\n",
"dataset[\"validation\"] = dataset[\"test\"]\n",
"del(dataset[\"test\"])\n",
"\n",
"classes = dataset[\"train\"].features[\"label\"].names\n",
"dataset = dataset.map(\n",
" lambda x: {\"text_label\": [classes[label] for label in x[\"label\"]]},\n",
" batched=True,\n",
" num_proc=1,\n",
" \n",
")\n",
"\n",
"dataset[\"train\"][0]"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "adf9608c",
"metadata": {},
"outputs": [
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "2ce088f4437d4e2c80c267332a5b84e5",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Downloading: 0%| | 0.00/792k [00:00<?, ?B/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "4e5f69b61f194220b39336e48edd2f9e",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Downloading: 0%| | 0.00/1.39M [00:00<?, ?B/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/sourab/transformers/src/transformers/models/t5/tokenization_t5_fast.py:156: FutureWarning: This tokenizer was incorrectly instantiated with a model max length of 512 which will be corrected in Transformers v5.\n",
"For now, this behavior is kept to avoid breaking backwards compatibility when padding/encoding with `truncation is True`.\n",
"- Be aware that you SHOULD NOT rely on t5-large automatically truncating your input to 512 when padding/encoding.\n",
"- If you want to encode/pad to sequences longer than 512 you can either instantiate this tokenizer with `model_max_length` or pass `max_length` when encoding/padding.\n",
"- To avoid this warning, please instantiate this tokenizer with `model_max_length` set to your preferred value.\n",
" warnings.warn(\n"
]
},
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "230c5631891e4ea8ac7a1b39f315a4f0",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Running tokenizer on dataset: 0%| | 0/3 [00:00<?, ?ba/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "b581e5677d2a45459ceb725534ed0891",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Running tokenizer on dataset: 0%| | 0/1 [00:00<?, ?ba/s]"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# data preprocessing\n",
"tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)\n",
"def preprocess_function(examples):\n",
" inputs = examples[text_column]\n",
" targets = examples[label_column]\n",
" model_inputs = tokenizer(inputs, max_length=max_length, padding=\"max_length\", truncation=True, return_tensors=\"pt\")\n",
" labels = tokenizer(targets, max_length=2, padding=\"max_length\", truncation=True, return_tensors=\"pt\")\n",
" labels = labels[\"input_ids\"]\n",
" labels[labels==tokenizer.pad_token_id] = -100\n",
" model_inputs[\"labels\"] = labels\n",
" return model_inputs\n",
"\n",
"processed_datasets = dataset.map(\n",
" preprocess_function,\n",
" batched=True,\n",
" num_proc=1,\n",
" remove_columns=dataset[\"train\"].column_names,\n",
" load_from_cache_file=False,\n",
" desc=\"Running tokenizer on dataset\",\n",
" )\n",
"\n",
"train_dataset = processed_datasets[\"train\"]\n",
"eval_dataset = processed_datasets[\"validation\"]\n",
"\n",
"train_dataloader = DataLoader(\n",
" train_dataset, shuffle=True, collate_fn=default_data_collator, batch_size=batch_size, pin_memory=True\n",
" )\n",
"eval_dataloader = DataLoader(eval_dataset, collate_fn=default_data_collator, batch_size=batch_size, pin_memory=True)\n",
"\n",
"\n",
"\n",
" "
]
},
{
"cell_type": "code",
"execution_count": 5,
"id": "f733a3c6",
"metadata": {},
"outputs": [],
"source": [
"# optimizer and lr scheduler\n",
"optimizer = torch.optim.AdamW(model.parameters(), lr=lr)\n",
"lr_scheduler = get_linear_schedule_with_warmup(\n",
" optimizer=optimizer,\n",
" num_warmup_steps=0,\n",
" num_training_steps=(len(train_dataloader) * num_epochs),\n",
")\n"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "6b3a4090",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|█████████████████████████████████████████████████████████████| 255/255 [00:20<00:00, 12.27it/s]\n",
"100%|███████████████████████████████████████████████████████████████| 29/29 [00:01<00:00, 17.32it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"epoch=0: train_ppl=tensor(2697769., device='cuda:0') train_epoch_loss=tensor(14.8079, device='cuda:0') eval_ppl=tensor(1.0089, device='cuda:0') eval_epoch_loss=tensor(0.0089, device='cuda:0')\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|█████████████████████████████████████████████████████████████| 255/255 [00:19<00:00, 12.75it/s]\n",
"100%|███████████████████████████████████████████████████████████████| 29/29 [00:01<00:00, 17.33it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"epoch=1: train_ppl=tensor(2.9475, device='cuda:0') train_epoch_loss=tensor(1.0809, device='cuda:0') eval_ppl=tensor(1.0072, device='cuda:0') eval_epoch_loss=tensor(0.0072, device='cuda:0')\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|█████████████████████████████████████████████████████████████| 255/255 [00:20<00:00, 12.71it/s]\n",
"100%|███████████████████████████████████████████████████████████████| 29/29 [00:01<00:00, 17.31it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"epoch=2: train_ppl=tensor(2.0588, device='cuda:0') train_epoch_loss=tensor(0.7221, device='cuda:0') eval_ppl=tensor(1.0055, device='cuda:0') eval_epoch_loss=tensor(0.0054, device='cuda:0')\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|█████████████████████████████████████████████████████████████| 255/255 [00:20<00:00, 12.70it/s]\n",
"100%|███████████████████████████████████████████████████████████████| 29/29 [00:01<00:00, 17.32it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"epoch=3: train_ppl=tensor(1.7939, device='cuda:0') train_epoch_loss=tensor(0.5844, device='cuda:0') eval_ppl=tensor(1.0063, device='cuda:0') eval_epoch_loss=tensor(0.0063, device='cuda:0')\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|█████████████████████████████████████████████████████████████| 255/255 [00:19<00:00, 13.01it/s]\n",
"100%|███████████████████████████████████████████████████████████████| 29/29 [00:01<00:00, 17.33it/s]"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"epoch=4: train_ppl=tensor(1.7740, device='cuda:0') train_epoch_loss=tensor(0.5732, device='cuda:0') eval_ppl=tensor(1.0062, device='cuda:0') eval_epoch_loss=tensor(0.0061, device='cuda:0')\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"\n"
]
}
],
"source": [
"# training and evaluation\n",
"model = model.to(device)\n",
"\n",
"for epoch in range(num_epochs):\n",
" model.train()\n",
" total_loss = 0\n",
" for step, batch in enumerate(tqdm(train_dataloader)):\n",
" batch = {k: v.to(device) for k, v in batch.items()}\n",
" outputs = model(**batch)\n",
" loss = outputs.loss\n",
" total_loss += loss.detach().float()\n",
" loss.backward()\n",
" optimizer.step()\n",
" lr_scheduler.step()\n",
" optimizer.zero_grad()\n",
"\n",
" model.eval()\n",
" eval_loss = 0\n",
" eval_preds = []\n",
" for step, batch in enumerate(tqdm(eval_dataloader)):\n",
" batch = {k: v.to(device) for k, v in batch.items()}\n",
" with torch.no_grad():\n",
" outputs = model(**batch)\n",
" loss = outputs.loss\n",
" eval_loss += loss.detach().float()\n",
" eval_preds.extend(tokenizer.batch_decode(torch.argmax(outputs.logits, -1).detach().cpu().numpy(), skip_special_tokens=True))\n",
"\n",
" eval_epoch_loss = eval_loss/len(train_dataloader)\n",
" eval_ppl = torch.exp(eval_epoch_loss)\n",
" train_epoch_loss = total_loss/len(eval_dataloader)\n",
" train_ppl = torch.exp(train_epoch_loss)\n",
" print(f\"{epoch=}: {train_ppl=} {train_epoch_loss=} {eval_ppl=} {eval_epoch_loss=}\")\n"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "6cafa67b",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"accuracy=96.47577092511013 % on the evaluation dataset\n",
"eval_preds[:10]=['neutral', 'neutral', 'neutral', 'negative', 'neutral', 'neutral', 'neutral', 'neutral', 'positive', 'positive']\n",
"dataset['validation']['text_label'][:10]=['neutral', 'neutral', 'neutral', 'negative', 'neutral', 'neutral', 'neutral', 'neutral', 'positive', 'positive']\n"
]
}
],
"source": [
"# print accuracy\n",
"correct =0\n",
"total = 0\n",
"for pred,true in zip(eval_preds, dataset[\"validation\"][\"text_label\"]):\n",
" if pred.strip()==true.strip():\n",
" correct+=1\n",
" total+=1 \n",
"accuracy = correct/total*100\n",
"print(f\"{accuracy=} % on the evaluation dataset\")\n",
"print(f\"{eval_preds[:10]=}\")\n",
"print(f\"{dataset['validation']['text_label'][:10]=}\")"
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "a8de6005",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{'prompt_embeddings': tensor([[-0.3165, -0.8389, 0.3262, ..., -1.5049, -1.6963, 0.3444],\n",
" [-1.8359, 1.1936, 1.0483, ..., 0.6197, -0.4452, 0.5844],\n",
" [-0.6027, 0.3246, -1.5601, ..., -0.3645, 0.2329, 0.3402],\n",
" ...,\n",
" [-1.9525, -0.5035, 0.8474, ..., 0.4793, -0.0789, -0.9305],\n",
" [-1.9741, 0.5242, -2.0594, ..., -0.7970, -0.4889, 2.7323],\n",
" [ 0.9355, -0.2714, 0.4610, ..., 0.2692, -1.5801, -1.6405]])}\n"
]
}
],
"source": [
"# saving model\n",
"state_dict = get_pet_model_state_dict(model)\n",
"torch.save(state_dict, checkpoint_name)\n",
"print(state_dict)"
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "bd20cd4c",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"3,8M\tfinancial_sentiment_analysis_prefix_tuning_v1.pt\r\n"
]
}
],
"source": [
"!du -h $checkpoint_name"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "76c2fc29",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3.10.5 64-bit",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.10.5 (v3.10.5:f377153967, Jun 6 2022, 12:36:10) [Clang 13.0.0 (clang-1300.0.29.30)]"
},
"vscode": {
"interpreter": {
"hash": "aee8b7b246df8f9039afb4144a1f6fd8d2ca17a180786b69acc140d282b71a49"
}
}
},
"nbformat": 4,
"nbformat_minor": 5
}