[feature] add initial version of anthropic dataset

model/reward/instructor/rank_datasets.py

@@ -237,3 +237,39 @@ class HFDataset(Dataset):
 class GPTJSynthetic(HFDataset):
     def __init__(self) -> None:
         super().__init__("Dahoas/synthetic-instruct-gptj-pairwise", "prompt", "chosen", "rejected", None, "train")
+
+
+class AnthropicRLHF(Dataset):
+    """
+    The data are described in the paper:
+    Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback.
+    If you find the data useful, please cite the paper.
+    The data format is very simple -- each line of the jsonl files contains a pair of texts,
+    one "chosen" and one "rejected".
+
+    """
+
+    def __init__(self, split="train", sep_token="<sep>") -> None:
+        super().__init__()
+        assert split in ("train", "test")
+        if sep_token is None:
+            sep_token = " . "
+        self.pairs = []
+        # using prompt as our index will allows us
+        # to add additional generated prompt later
+        major_split = split if "train" == split else "test"
+        dataset = load_dataset("Anthropic/hh-rlhf")[major_split]
+        for data in dataset:
+            prompt, pos_postfix = data["chosen"].split("Assistant:", maxsplit=1)
+            pos_postfix = pos_postfix.replace("\n\nHuman: ", sep_token).replace("\n\nAssistant: ", sep_token)
+            _, neg_postfix = data["rejected"].split("Assistant:", maxsplit=1)
+            neg_postfix = neg_postfix.replace("\n\nHuman: ", sep_token).replace("\n\nAssistant: ", sep_token)
+            self.pairs.append((prompt, (pos_postfix.strip(), neg_postfix.strip())))
+
+    def __len__(self):
+        return len(self.pairs)
+
+    def __getitem__(self, index):
+        context, pair = self.pairs[index]
+
+        return context, [pair]

model/reward/instructor/tests/test_dataset.py

@@ -1,5 +1,5 @@
 from experimental_dataset import DataCollatorForSummaryScore, HFSummaryQuality
-from rank_datasets import DataCollatorForPairRank, GPTJSynthetic, HFSummary, WebGPT
+from rank_datasets import AnthropicRLHF, DataCollatorForPairRank, GPTJSynthetic, HFSummary, WebGPT
 from torch.utils.data import DataLoader
 from transformers import AutoTokenizer
 
@@ -25,6 +25,16 @@ def test_webgpt():
         print(batch["input_ids"].shape)
 
 
+def test_anthropic_rlhf():
+
+    tokenizer = AutoTokenizer.from_pretrained("bigscience/mt0-large")
+    collate_fn = DataCollatorForPairRank(tokenizer, max_length=200)
+    dataset = AnthropicRLHF("test", sep_token=tokenizer.sep_token)
+    dataloader = DataLoader(dataset, collate_fn=collate_fn, batch_size=32)
+    for batch in dataloader:
+        print(batch["input_ids"].shape)
+
+
 def test_hf_summary_quality():
 
     tokenizer = AutoTokenizer.from_pretrained("bigscience/mt0-large")

model/reward/instructor/trainer.py

@@ -155,7 +155,7 @@ if __name__ == "__main__":
     )
 
     tokenizer = get_tokenizer(training_conf["tokenizer_name"])
-    train, evals = get_datasets(training_conf["datasets"])
+    train, evals = get_datasets(training_conf["datasets"], tokenizer)
     if "rankgen" in model_name:
         collate_fn = RankGenCollator(tokenizer, max_length=training_conf["max_length"])
     else:

model/reward/instructor/utils.py

@@ -99,8 +99,8 @@ def argument_parsing(parser):
     return params
 
 
-def get_datasets(dataset_list: List[AnyStr]):
-    from rank_datasets import GPTJSynthetic, HFSummary, WebGPT
+def get_datasets(dataset_list: List[AnyStr], tokenizer):
+    from rank_datasets import AnthropicRLHF, GPTJSynthetic, HFSummary, WebGPT
     from torch.utils.data import ConcatDataset
 
     train_datasets, evals = [], {}
@@ -121,6 +121,11 @@ def get_datasets(dataset_list: List[AnyStr]):
             train, eval = train_val_dataset(dataset, 0.1)
             train_datasets.append(train)
             evals["gptsynthetic"] = eval
+        elif "anthropic_rlhf" == dataset_name:
+            train = AnthropicRLHF("train", tokenizer.sep_token)
+            eval = AnthropicRLHF("test", tokenizer.sep_token)
+            train_datasets.append(train)
+            evals["anthropic_rlhf"] = eval
     train = ConcatDataset(train_datasets)
     return train, evals