Skip to content

dataset

BaseTrainDataset

Bases: Dataset, ABC

Base class for chat reinforcement learning datasets with common functionality

Source code in rm_gallery/core/train/dataset.py
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
class BaseTrainDataset(Dataset, ABC):
    """Base class for chat reinforcement learning datasets with common functionality"""

    def __init__(
        self,
        data_files: Union[str, List[str]],
        tokenizer: PreTrainedTokenizer,
        config: DictConfig,
        processor=None,  # keep for backward compatibility, but not used
    ):
        # initialize basic attributes
        self.data_files = self._normalize_data_files(data_files)
        self.original_data_files = copy.deepcopy(self.data_files)
        self.tokenizer = tokenizer
        self.config = config

        # load config settings
        self._load_config()

        # load and process data
        self._load_dataset()

    def _normalize_data_files(self, data_files):
        """Convert data files to list format"""
        if not isinstance(data_files, (List, ListConfig)):
            data_files = [data_files]
        return copy.deepcopy(data_files)

    def _load_config(self):
        """Load config parameters - can be overridden by subclasses"""
        self.cache_dir = os.path.expanduser(
            self.config.get("cache_dir", "~/.cache/verl/rlhf")
        )
        self.prompt_key = self.config.get("prompt_key", "prompt")
        self.max_prompt_length = self.config.get("max_prompt_length", 1024)
        self.return_raw_chat = self.config.get("return_raw_chat", False)
        self.truncation = self.config.get("truncation", "error")
        self.filter_overlong_prompts = self.config.get("filter_overlong_prompts", True)
        self.num_workers = min(
            self.config.get(
                "filter_overlong_prompts_workers", max(1, os.cpu_count() // 4)
            ),
            os.cpu_count(),
        )
        self.serialize_dataset = False

    def _download_files(self):
        """Download files to local cache"""
        from verl.utils.fs import copy_to_local

        for i, file in enumerate(self.data_files):
            self.data_files[i] = copy_to_local(src=file, cache_dir=self.cache_dir)

    def _load_dataset(self):
        """Load and process dataset"""
        self._download_files()

        # Load parquet files
        dataframes = []
        for file in self.data_files:
            df = datasets.load_dataset("parquet", data_files=file)["train"]
            dataframes.append(df)

        self.dataframe = datasets.concatenate_datasets(dataframes)
        print(f"dataset length: {len(self.dataframe)}")

        # Filter overlong prompts if enabled

        if self.filter_overlong_prompts:
            self._filter_long_prompts()

    def _filter_long_prompts(self):
        """Filter out overlong prompts using the same logic as runtime processing"""

        def is_prompt_valid(doc):
            try:
                # Use the same logic as runtime processing
                messages = self._build_messages(doc)
                raw_prompt = self._apply_chat_template(messages)
                raw_prompt_ids = self.tokenizer.encode(
                    raw_prompt, add_special_tokens=False
                )
                return len(raw_prompt_ids) <= self.max_prompt_length
            except Exception as e:
                print(f"Error processing sample during filtering: {e}")
                return False

        print("Starting prompt length filtering...")
        self.dataframe = self.dataframe.filter(
            is_prompt_valid,
            num_proc=self.num_workers,
            desc=f"filter out prompts longer than {self.max_prompt_length} tokens",
        )
        print(f"filtered dataset length: {len(self.dataframe)}")

    @abstractmethod
    def _build_messages(self, example: Dict[str, Any]) -> List[Dict[str, str]]:
        """Build chat messages from example - must be implemented by subclasses"""
        pass

    @abstractmethod
    def _apply_chat_template(self, messages: List[Dict[str, str]]) -> str:
        """Apply chat template - can be overridden by subclasses"""
        pass

    @abstractmethod
    def _extract_ground_truth(self, row_dict: Dict[str, Any]) -> str:
        """Extract ground truth from row data - must be implemented by subclasses"""
        pass

    @abstractmethod
    def _get_data_source(self, row_dict: Dict[str, Any]) -> str:
        """Get data source - can be overridden by subclasses"""
        pass

    def __getitem__(self, item):
        """Get an item from dataset"""
        row_dict = dict(self.dataframe[item])
        messages = self._build_messages(row_dict)
        raw_prompt = self._apply_chat_template(messages)

        # Tokenize
        model_inputs = self.tokenizer(
            raw_prompt, return_tensors="pt", add_special_tokens=False
        )
        input_ids = model_inputs["input_ids"]
        attention_mask = model_inputs["attention_mask"]

        # Postprocess
        input_ids, attention_mask = verl_F.postprocess_data(
            input_ids=input_ids,
            attention_mask=attention_mask,
            max_length=self.max_prompt_length,
            pad_token_id=self.tokenizer.pad_token_id,
            left_pad=True,
            truncation=self.truncation,
        )

        # Compute position ids
        position_ids = compute_position_id_with_mask(attention_mask)

        # Prepare raw prompt ids
        raw_prompt_ids = self.tokenizer.encode(raw_prompt, add_special_tokens=False)
        if len(raw_prompt_ids) > self.max_prompt_length:
            if self.truncation == "left":
                raw_prompt_ids = raw_prompt_ids[-self.max_prompt_length :]
            elif self.truncation == "right":
                raw_prompt_ids = raw_prompt_ids[: self.max_prompt_length]
            elif self.truncation == "error":
                raise RuntimeError(
                    f"prompt length {len(raw_prompt_ids)} exceeds {self.max_prompt_length}"
                )

        # Build result
        result = {
            "input_ids": input_ids[0],
            "attention_mask": attention_mask[0],
            "position_ids": position_ids[0],
            "raw_prompt_ids": raw_prompt_ids,
            "index": row_dict.get("index", item),
            "extra_info": copy.deepcopy(row_dict),
            "reward_model": {"ground_truth": self._extract_ground_truth(row_dict)},
            "data_source": self._get_data_source(row_dict),
        }

        if self.return_raw_chat:
            result["raw_prompt"] = messages

        return result

    def __len__(self):
        return len(self.dataframe)

    def resume_dataset_state(self):
        """Resume dataset state for checkpoint"""
        self.serialize_dataset = not hasattr(self, "original_data_files")
        if not self.serialize_dataset:
            self.data_files = copy.deepcopy(self.original_data_files)
            self._load_dataset()
        else:
            print(
                "use old dataset loader checkpoint file, it is recommended to train from scratch"
            )

    def __getstate__(self):
        """Get state for serialization"""
        if not self.serialize_dataset:
            state = self.__dict__.copy()
            if "dataframe" in state:
                del state["dataframe"]
            return state
        return self.__dict__.copy()

__getitem__(item)

Get an item from dataset

Source code in rm_gallery/core/train/dataset.py
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
def __getitem__(self, item):
    """Get an item from dataset"""
    row_dict = dict(self.dataframe[item])
    messages = self._build_messages(row_dict)
    raw_prompt = self._apply_chat_template(messages)

    # Tokenize
    model_inputs = self.tokenizer(
        raw_prompt, return_tensors="pt", add_special_tokens=False
    )
    input_ids = model_inputs["input_ids"]
    attention_mask = model_inputs["attention_mask"]

    # Postprocess
    input_ids, attention_mask = verl_F.postprocess_data(
        input_ids=input_ids,
        attention_mask=attention_mask,
        max_length=self.max_prompt_length,
        pad_token_id=self.tokenizer.pad_token_id,
        left_pad=True,
        truncation=self.truncation,
    )

    # Compute position ids
    position_ids = compute_position_id_with_mask(attention_mask)

    # Prepare raw prompt ids
    raw_prompt_ids = self.tokenizer.encode(raw_prompt, add_special_tokens=False)
    if len(raw_prompt_ids) > self.max_prompt_length:
        if self.truncation == "left":
            raw_prompt_ids = raw_prompt_ids[-self.max_prompt_length :]
        elif self.truncation == "right":
            raw_prompt_ids = raw_prompt_ids[: self.max_prompt_length]
        elif self.truncation == "error":
            raise RuntimeError(
                f"prompt length {len(raw_prompt_ids)} exceeds {self.max_prompt_length}"
            )

    # Build result
    result = {
        "input_ids": input_ids[0],
        "attention_mask": attention_mask[0],
        "position_ids": position_ids[0],
        "raw_prompt_ids": raw_prompt_ids,
        "index": row_dict.get("index", item),
        "extra_info": copy.deepcopy(row_dict),
        "reward_model": {"ground_truth": self._extract_ground_truth(row_dict)},
        "data_source": self._get_data_source(row_dict),
    }

    if self.return_raw_chat:
        result["raw_prompt"] = messages

    return result

__getstate__()

Get state for serialization

Source code in rm_gallery/core/train/dataset.py
212
213
214
215
216
217
218
219
def __getstate__(self):
    """Get state for serialization"""
    if not self.serialize_dataset:
        state = self.__dict__.copy()
        if "dataframe" in state:
            del state["dataframe"]
        return state
    return self.__dict__.copy()

resume_dataset_state()

Resume dataset state for checkpoint

Source code in rm_gallery/core/train/dataset.py
201
202
203
204
205
206
207
208
209
210
def resume_dataset_state(self):
    """Resume dataset state for checkpoint"""
    self.serialize_dataset = not hasattr(self, "original_data_files")
    if not self.serialize_dataset:
        self.data_files = copy.deepcopy(self.original_data_files)
        self._load_dataset()
    else:
        print(
            "use old dataset loader checkpoint file, it is recommended to train from scratch"
        )