Update Dockerfile, model.properties, requirements.txt, LICENSE.txt, whl.zip,...

Update Dockerfile, model.properties, requirements.txt, LICENSE.txt, whl.zip, doc/training_loss.png, doc/DeepSeekMath Pushing the Limits of Mathematical.pdf, finetune/multi_node.sh, finetune/single_node.sh, finetune/scripts/cal_lr.py, finetune/scripts/cal_ppl.py, finetune/scripts/cal_flops.py, finetune/scripts/length_cdf.py, finetune/scripts/llama_pro.py, finetune/scripts/llamafy_baichuan2.py, finetune/scripts/loftq_init.py, finetune/scripts/llamafy_qwen.py, finetune/data/dataset_info.json, finetune/data/mllm_demo.json, finetune/data/README.md, finetune/data/identity.json, finetune/data/README_zh.md, finetune/data/alpaca_zh_demo.json, finetune/data/glaive_toolcall_zh_demo.json, finetune/data/glaive_toolcall_en_demo.json, finetune/data/c4_demo.json, finetune/data/alpaca_en_demo.json, finetune/data/dpo_zh_demo.json, finetune/data/kto_en_demo.json, finetune/data/wiki_demo.txt, finetune/data/dpo_en_demo.json, finetune/src/api.py, finetune/src/train.py, finetune/src/webui.py, finetune/src/llmfactory/__init__.py, finetune/src/llmfactory/cli.py, finetune/src/llmfactory/api/__init__.py, finetune/src/llmfactory/api/protocol.py, finetune/src/llmfactory/api/chat.py, finetune/src/llmfactory/api/common.py, finetune/src/llmfactory/api/app.py, finetune/src/llmfactory/chat/__init__.py, finetune/src/llmfactory/chat/base_engine.py, finetune/src/llmfactory/chat/chat_model.py, finetune/src/llmfactory/chat/vllm_engine.py, finetune/src/llmfactory/chat/hf_engine.py, finetune/src/llmfactory/data/__init__.py, finetune/src/llmfactory/data/template.py, finetune/src/llmfactory/data/aligner.py, finetune/src/llmfactory/data/preprocess.py, finetune/src/llmfactory/data/loader.py, finetune/src/llmfactory/data/collator.py, finetune/src/llmfactory/data/parser.py, finetune/src/llmfactory/data/formatter.py, finetune/src/llmfactory/data/utils.py, finetune/src/llmfactory/eval/__init__.py, finetune/src/llmfactory/eval/template.py, finetune/src/llmfactory/eval/evaluator.py, finetune/src/llmfactory/extras/__init__.py, finetune/src/llmfactory/extras/ploting.py, finetune/src/llmfactory/extras/packages.py, finetune/src/llmfactory/extras/constants.py, finetune/src/llmfactory/extras/misc.py, finetune/src/llmfactory/extras/logging.py, finetune/src/llmfactory/extras/callbacks.py, finetune/src/llmfactory/hparams/__init__.py, finetune/src/llmfactory/hparams/data_args.py, finetune/src/llmfactory/hparams/evaluation_args.py, finetune/src/llmfactory/hparams/finetuning_args.py, finetune/src/llmfactory/hparams/generating_args.py, finetune/src/llmfactory/hparams/parser.py, finetune/src/llmfactory/hparams/model_args.py, finetune/src/llmfactory/model/__init__.py, finetune/src/llmfactory/model/adapter.py, finetune/src/llmfactory/model/patcher.py, finetune/src/llmfactory/model/loader.py, finetune/src/llmfactory/model/utils/__init__.py, finetune/src/llmfactory/model/utils/attention.py, finetune/src/llmfactory/model/utils/checkpointing.py, finetune/src/llmfactory/model/utils/longlora.py, finetune/src/llmfactory/model/utils/embedding.py, finetune/src/llmfactory/model/utils/unsloth.py, finetune/src/llmfactory/model/utils/moe.py, finetune/src/llmfactory/model/utils/misc.py, finetune/src/llmfactory/model/utils/rope.py, finetune/src/llmfactory/model/utils/quantization.py, finetune/src/llmfactory/model/utils/mod.py, finetune/src/llmfactory/model/utils/valuehead.py, finetune/src/llmfactory/model/utils/visual.py, finetune/src/llmfactory/train/__init__.py, finetune/src/llmfactory/train/tuner.py, finetune/src/llmfactory/train/utils.py, finetune/src/llmfactory/train/dpo/__init__.py, finetune/src/llmfactory/train/dpo/workflow.py, finetune/src/llmfactory/train/dpo/trainer.py, finetune/src/llmfactory/train/kto/__init__.py, finetune/src/llmfactory/train/kto/trainer.py, finetune/src/llmfactory/train/kto/workflow.py, finetune/src/llmfactory/train/orpo/__init__.py, finetune/src/llmfactory/train/orpo/workflow.py, finetune/src/llmfactory/train/orpo/trainer.py, finetune/src/llmfactory/train/ppo/__init__.py, finetune/src/llmfactory/train/ppo/workflow.py, finetune/src/llmfactory/train/ppo/trainer.py, finetune/src/llmfactory/train/ppo/utils.py, finetune/src/llmfactory/train/pt/__init__.py, finetune/src/llmfactory/train/pt/trainer.py, finetune/src/llmfactory/train/pt/workflow.py, finetune/src/llmfactory/train/rm/__init__.py, finetune/src/llmfactory/train/rm/trainer.py, finetune/src/llmfactory/train/rm/workflow.py, finetune/src/llmfactory/train/rm/metric.py, finetune/src/llmfactory/train/sft/__init__.py, finetune/src/llmfactory/train/sft/workflow.py, finetune/src/llmfactory/train/sft/metric.py, finetune/src/llmfactory/train/sft/trainer.py, finetune/src/llmfactory/webui/__init__.py, finetune/src/llmfactory/webui/chatter.py, finetune/src/llmfactory/webui/engine.py, finetune/src/llmfactory/webui/interface.py, finetune/src/llmfactory/webui/css.py, finetune/src/llmfactory/webui/manager.py, finetune/src/llmfactory/webui/utils.py, finetune/src/llmfactory/webui/common.py, finetune/src/llmfactory/webui/locales.py, finetune/src/llmfactory/webui/runner.py, finetune/src/llmfactory/webui/components/chatbot.py, finetune/src/llmfactory/webui/components/__init__.py, finetune/src/llmfactory/webui/components/train.py, finetune/src/llmfactory/webui/components/top.py, finetune/src/llmfactory/webui/components/eval.py, finetune/src/llmfactory/webui/components/export.py, finetune/src/llmfactory/webui/components/infer.py, finetune/src/llmfactory/webui/components/data.py, inference/7B_single_dcu.py files

finetune/src/llmfactory/cli.py

+import sys
+from enum import Enum, unique
+
+from .api.app import run_api
+from .chat.chat_model import run_chat
+from .eval.evaluator import run_eval
+from .train.tuner import export_model, run_exp
+from .webui.interface import run_web_demo, run_web_ui
+
+
+USAGE = (
+    "-" * 70
+    + "\n"
+    + "| Usage:                                                             |\n"
+    + "|   llamafactory-cli api -h: launch an OpenAI-style API server       |\n"
+    + "|   llamafactory-cli chat -h: launch a chat interface in CLI         |\n"
+    + "|   llamafactory-cli eval -h: evaluate models                        |\n"
+    + "|   llamafactory-cli export -h: merge LoRA adapters and export model |\n"
+    + "|   llamafactory-cli train -h: train models                          |\n"
+    + "|   llamafactory-cli webchat -h: launch a chat interface in Web UI   |\n"
+    + "|   llamafactory-cli webui: launch LlamaBoard                        |\n"
+    + "|   llamafactory-cli version: show version info                      |\n"
+    + "-" * 70
+)
+
+VERSION = "0.7.2.dev0"
+
+WELCOME = (
+    "-" * 58
+    + "\n"
+    + "| Welcome to LLaMA Factory, version {}".format(VERSION)
+    + " " * (21 - len(VERSION))
+    + "|\n|"
+    + " " * 56
+    + "|\n"
+    + "| Project page: https://github.com/hiyouga/LLaMA-Factory |\n"
+    + "-" * 58
+)
+
+
+@unique
+class Command(str, Enum):
+    API = "api"
+    CHAT = "chat"
+    EVAL = "eval"
+    EXPORT = "export"
+    TRAIN = "train"
+    WEBDEMO = "webchat"
+    WEBUI = "webui"
+    VER = "version"
+    HELP = "help"
+
+
+def main():
+    command = sys.argv.pop(1)
+    if command == Command.API:
+        run_api()
+    elif command == Command.CHAT:
+        run_chat()
+    elif command == Command.EVAL:
+        run_eval()
+    elif command == Command.EXPORT:
+        export_model()
+    elif command == Command.TRAIN:
+        run_exp()
+    elif command == Command.WEBDEMO:
+        run_web_demo()
+    elif command == Command.WEBUI:
+        run_web_ui()
+    elif command == Command.VER:
+        print(WELCOME)
+    elif command == Command.HELP:
+        print(USAGE)
+    else:
+        raise NotImplementedError("Unknown command: {}".format(command))

finetune/src/llmfactory/data/__init__.py

+from .collator import KTODataCollatorWithPadding, PairwiseDataCollatorWithPadding
+from .loader import get_dataset
+from .template import Template, get_template_and_fix_tokenizer, templates
+from .utils import Role, split_dataset
+
+
+__all__ = [
+    "KTODataCollatorWithPadding",
+    "PairwiseDataCollatorWithPadding",
+    "get_dataset",
+    "Template",
+    "get_template_and_fix_tokenizer",
+    "templates",
+    "Role",
+    "split_dataset",
+]

finetune/src/llmfactory/data/aligner.py

+import os
+from functools import partial
+from typing import TYPE_CHECKING, Any, Dict, List, Union
+
+from datasets import Features
+
+from ..extras.logging import get_logger
+from .utils import Role
+
+
+if TYPE_CHECKING:
+    from datasets import Dataset, IterableDataset
+
+    from ..hparams import DataArguments
+    from .parser import DatasetAttr
+
+
+logger = get_logger(__name__)
+
+
+def _convert_images(images: List[Any], dataset_attr: "DatasetAttr", data_args: "DataArguments") -> List[Any]:
+    r"""
+    Optionally concatenates image path to dataset dir when loading from local disk.
+    """
+    outputs = []
+    if dataset_attr.load_from in ["script", "file"]:
+        for image in images:
+            if isinstance(image, str) and os.path.isfile(os.path.join(data_args.dataset_dir, image)):
+                outputs.append(os.path.join(data_args.dataset_dir, image))
+            else:
+                outputs.append(image)
+
+    return outputs
+
+
+def convert_alpaca(
+    examples: Dict[str, List[Any]], dataset_attr: "DatasetAttr", data_args: "DataArguments"
+) -> Dict[str, List[Any]]:
+    r"""
+    Converts alpaca format dataset to the standard format.
+    """
+    outputs = {"prompt": [], "response": [], "system": [], "tools": [], "images": []}
+    convert_images = partial(_convert_images, dataset_attr=dataset_attr, data_args=data_args)
+    for i in range(len(examples[dataset_attr.prompt])):
+        prompt = []
+        if dataset_attr.history and isinstance(examples[dataset_attr.history][i], list):
+            for old_prompt, old_response in examples[dataset_attr.history][i]:
+                prompt.append({"role": Role.USER.value, "content": old_prompt})
+                prompt.append({"role": Role.ASSISTANT.value, "content": old_response})
+
+        content = []
+        if dataset_attr.prompt and examples[dataset_attr.prompt][i]:
+            content.append(examples[dataset_attr.prompt][i])
+
+        if dataset_attr.query and examples[dataset_attr.query][i]:
+            content.append(examples[dataset_attr.query][i])
+
+        prompt.append({"role": Role.USER.value, "content": "\n".join(content)})  # "prompt\nquery"
+
+        if dataset_attr.kto_tag and isinstance(examples[dataset_attr.kto_tag][i], bool):  # kto example
+            response = [{"role": Role.ASSISTANT.value, "content": examples[dataset_attr.response][i]}]
+            if examples[dataset_attr.kto_tag][i]:
+                response = response + [{"role": Role.ASSISTANT.value, "content": ""}]
+            else:
+                response = [{"role": Role.ASSISTANT.value, "content": ""}] + response
+        elif (
+            dataset_attr.ranking
+            and isinstance(examples[dataset_attr.chosen][i], str)
+            and isinstance(examples[dataset_attr.rejected][i], str)
+        ):  # pairwise example
+            response = [
+                {"role": Role.ASSISTANT.value, "content": examples[dataset_attr.chosen][i]},
+                {"role": Role.ASSISTANT.value, "content": examples[dataset_attr.rejected][i]},
+            ]
+        elif dataset_attr.response and isinstance(examples[dataset_attr.response][i], str):  # normal example
+            response = [{"role": Role.ASSISTANT.value, "content": examples[dataset_attr.response][i]}]
+        else:  # unsupervised
+            response = []
+
+        outputs["prompt"].append(prompt)
+        outputs["response"].append(response)
+        outputs["system"].append(examples[dataset_attr.system][i] if dataset_attr.system else "")
+        outputs["tools"].append(examples[dataset_attr.tools][i] if dataset_attr.tools else "")
+        outputs["images"].append(convert_images(examples[dataset_attr.images][i]) if dataset_attr.images else [])
+
+    return outputs
+
+
+def convert_sharegpt(
+    examples: Dict[str, List[Any]], dataset_attr: "DatasetAttr", data_args: "DataArguments"
+) -> Dict[str, List[Any]]:
+    r"""
+    Converts sharegpt format dataset to the standard format.
+    """
+    outputs = {"prompt": [], "response": [], "system": [], "tools": [], "images": []}
+    convert_images = partial(_convert_images, dataset_attr=dataset_attr, data_args=data_args)
+    tag_mapping = {
+        dataset_attr.user_tag: Role.USER.value,
+        dataset_attr.assistant_tag: Role.ASSISTANT.value,
+        dataset_attr.observation_tag: Role.OBSERVATION.value,
+        dataset_attr.function_tag: Role.FUNCTION.value,
+        dataset_attr.system_tag: Role.SYSTEM.value,
+    }
+    odd_tags = (dataset_attr.user_tag, dataset_attr.observation_tag)
+    even_tags = (dataset_attr.assistant_tag, dataset_attr.function_tag)
+    accept_tags = (odd_tags, even_tags)
+    for i, messages in enumerate(examples[dataset_attr.messages]):
+        if dataset_attr.system_tag and messages[0][dataset_attr.role_tag] == dataset_attr.system_tag:
+            system = messages[0][dataset_attr.content_tag]
+            messages = messages[1:]
+        else:
+            system = examples[dataset_attr.system][i] if dataset_attr.system else ""
+
+        if len(messages) == 0:
+            continue
+
+        aligned_messages = []
+        broken_data = False
+        for turn_idx, message in enumerate(messages):
+            if message[dataset_attr.role_tag] not in accept_tags[turn_idx % 2]:
+                logger.warning("Invalid role tag in {}.".format(messages))
+                broken_data = True
+
+            aligned_messages.append(
+                {"role": tag_mapping[message[dataset_attr.role_tag]], "content": message[dataset_attr.content_tag]}
+            )
+
+        if (not dataset_attr.ranking and len(aligned_messages) % 2 != 0) or (
+            dataset_attr.ranking and len(aligned_messages) % 2 == 0
+        ):
+            logger.warning("Invalid message count in {}.".format(messages))
+            broken_data = True
+
+        if dataset_attr.kto_tag and isinstance(examples[dataset_attr.kto_tag][i], bool):  # kto example
+            prompt = aligned_messages[:-1]
+            response = aligned_messages[-1:]
+            if examples[dataset_attr.kto_tag][i]:
+                response = response + [{"role": Role.ASSISTANT.value, "content": ""}]
+            else:
+                response = [{"role": Role.ASSISTANT.value, "content": ""}] + response
+        elif (
+            dataset_attr.ranking
+            and isinstance(examples[dataset_attr.chosen][i], dict)
+            and isinstance(examples[dataset_attr.rejected][i], dict)
+        ):  # pairwise example
+            chosen = examples[dataset_attr.chosen][i]
+            rejected = examples[dataset_attr.rejected][i]
+            if (
+                chosen[dataset_attr.role_tag] not in accept_tags[-1]
+                or rejected[dataset_attr.role_tag] not in accept_tags[-1]
+            ):
+                logger.warning("Invalid role tag in {}.".format([chosen, rejected]))
+                broken_data = True
+
+            prompt = aligned_messages
+            response = [
+                {"role": tag_mapping[chosen[dataset_attr.role_tag]], "content": chosen[dataset_attr.content_tag]},
+                {"role": tag_mapping[rejected[dataset_attr.role_tag]], "content": rejected[dataset_attr.content_tag]},
+            ]
+        else:  # normal example
+            prompt = aligned_messages[:-1]
+            response = aligned_messages[-1:]
+
+        if broken_data:
+            logger.warning("Skipping this abnormal example.")
+            continue
+
+        outputs["prompt"].append(prompt)
+        outputs["response"].append(response)
+        outputs["system"].append(system)
+        outputs["tools"].append(examples[dataset_attr.tools][i] if dataset_attr.tools else "")
+        outputs["images"].append(convert_images(examples[dataset_attr.images][i]) if dataset_attr.images else [])
+
+    return outputs
+
+
+def align_dataset(
+    dataset: Union["Dataset", "IterableDataset"], dataset_attr: "DatasetAttr", data_args: "DataArguments"
+) -> Union["Dataset", "IterableDataset"]:
+    r"""
+    Aligned dataset:
+        prompt: [{"role": "user", "content": "..."}] * (2T - 1)
+        response: [{"role": "assistant", "content": "..."}] * N (N > 1 for ranking dataset)
+        system: "..."
+        tools: "...",
+        images: [],
+    """
+    if dataset_attr.formatting == "alpaca":
+        convert_func = partial(convert_alpaca, dataset_attr=dataset_attr, data_args=data_args)
+    else:
+        convert_func = partial(convert_sharegpt, dataset_attr=dataset_attr, data_args=data_args)
+
+    column_names = list(next(iter(dataset)).keys())
+    features = Features.from_dict(
+        {
+            "prompt": [
+                {"role": {"dtype": "string", "_type": "Value"}, "content": {"dtype": "string", "_type": "Value"}}
+            ],
+            "response": [
+                {"role": {"dtype": "string", "_type": "Value"}, "content": {"dtype": "string", "_type": "Value"}}
+            ],
+            "system": {"dtype": "string", "_type": "Value"},
+            "tools": {"dtype": "string", "_type": "Value"},
+            "images": [{"_type": "Image"}],
+        }
+    )
+    kwargs = {}
+    if not data_args.streaming:
+        kwargs = dict(
+            num_proc=data_args.preprocessing_num_workers,
+            load_from_cache_file=(not data_args.overwrite_cache),
+            desc="Converting format of dataset",
+        )
+
+    return dataset.map(
+        convert_func,
+        batched=True,
+        remove_columns=column_names,
+        features=features,
+        **kwargs,
+    )

finetune/src/llmfactory/data/collator.py

+from dataclasses import dataclass
+from typing import Any, Dict, List, Sequence, Tuple
+
+import torch
+from transformers import DataCollatorForSeq2Seq
+
+
+@dataclass
+class PairwiseDataCollatorWithPadding(DataCollatorForSeq2Seq):
+    r"""
+    Data collator for pairwise data.
+    """
+
+    def _pad_labels(self, batch: torch.Tensor, positions: List[Tuple[int, int]]) -> torch.Tensor:
+        r"""
+        Masks out the input ids except for the responses.
+        """
+        padded_labels = []
+        for feature, (prompt_len, answer_len) in zip(batch, positions):
+            if self.tokenizer.padding_side == "left":
+                start, end = feature.size(0) - answer_len, feature.size(0)
+            else:
+                start, end = prompt_len, prompt_len + answer_len
+            padded_tensor = self.label_pad_token_id * torch.ones_like(feature)
+            padded_tensor[start:end] = feature[start:end]
+            padded_labels.append(padded_tensor)
+        return torch.stack(padded_labels, dim=0).contiguous()  # in contiguous memory
+
+    def __call__(self, features: Sequence[Dict[str, Any]]) -> Dict[str, torch.Tensor]:
+        r"""
+        Pads batched data to the longest sequence in the batch.
+
+        We generate 2 * n examples where the first n examples represent chosen examples and
+        the last n examples represent rejected examples.
+        """
+        concatenated_features = []
+        label_positions = []
+        for key in ("chosen_ids", "rejected_ids"):
+            for feature in features:
+                prompt_len, answer_len = len(feature["prompt_ids"]), len(feature[key])
+                concatenated_features.append(
+                    {
+                        "input_ids": feature["prompt_ids"] + feature[key],
+                        "attention_mask": [1] * (prompt_len + answer_len),
+                    }
+                )
+                label_positions.append((prompt_len, answer_len))
+
+        batch = super().__call__(concatenated_features)
+        batch["labels"] = self._pad_labels(batch["input_ids"], label_positions)
+        return batch
+
+
+@dataclass
+class KTODataCollatorWithPadding(DataCollatorForSeq2Seq):
+    r"""
+    Data collator for KTO data.
+    """
+
+    def __call__(self, features: Sequence[Dict[str, Any]]) -> Dict[str, torch.Tensor]:
+        target_features = []
+        kl_features = []
+        kto_tags = []
+        for feature in features:
+            target_features.append(
+                {
+                    "input_ids": feature["input_ids"],
+                    "attention_mask": feature["attention_mask"],
+                    "labels": feature["labels"],
+                }
+            )
+            kl_features.append(
+                {
+                    "input_ids": feature["kl_input_ids"],
+                    "attention_mask": feature["kl_attention_mask"],
+                    "labels": feature["kl_labels"],
+                }
+            )
+            kto_tags.append(feature["kto_tags"])
+
+        batch = super().__call__(target_features)
+        kl_batch = super().__call__(kl_features)
+        batch["kl_input_ids"] = kl_batch["input_ids"]
+        batch["kl_attention_mask"] = kl_batch["attention_mask"]
+        batch["kl_labels"] = kl_batch["labels"]
+        batch["kto_tags"] = torch.tensor(kto_tags)
+        return batch

finetune/src/llmfactory/data/formatter.py

+import json
+import re
+from abc import ABC, abstractmethod
+from dataclasses import dataclass, field
+from typing import Any, Dict, List, Literal, Optional, Sequence, Set, Tuple, Union
+
+
+SLOTS = Sequence[Union[str, Set[str], Dict[str, str]]]
+
+
+JSON_FORMAT_PROMPT = (
+    """, in a JSON format representing the kwargs (e.g. ```{"input": "hello world", "num_beams": 5}```)"""
+)
+
+
+TOOL_SYSTEM_PROMPT = (
+    "You have access to the following tools:\n{tool_text}"
+    "Use the following format if using a tool:\n"
+    "```\n"
+    "Action: tool name (one of [{tool_names}]).\n"
+    "Action Input: the input to the tool{format_prompt}.\n"
+    "```\n"
+)
+
+
+def default_tool_formatter(tools: List[Dict[str, Any]]) -> str:
+    tool_text = ""
+    tool_names = []
+    for tool in tools:
+        param_text = ""
+        for name, param in tool["parameters"]["properties"].items():
+            required = ", required" if name in tool["parameters"].get("required", []) else ""
+            enum = ", should be one of [{}]".format(", ".join(param["enum"])) if param.get("enum", None) else ""
+            items = (
+                ", where each item should be {}".format(param["items"].get("type", "")) if param.get("items") else ""
+            )
+            param_text += "  - {name} ({type}{required}): {desc}{enum}{items}\n".format(
+                name=name,
+                type=param.get("type", ""),
+                required=required,
+                desc=param.get("description", ""),
+                enum=enum,
+                items=items,
+            )
+
+        tool_text += "> Tool Name: {name}\nTool Description: {desc}\nTool Args:\n{args}\n".format(
+            name=tool["name"], desc=tool.get("description", ""), args=param_text
+        )
+        tool_names.append(tool["name"])
+
+    return TOOL_SYSTEM_PROMPT.format(
+        tool_text=tool_text, tool_names=", ".join(tool_names), format_prompt=JSON_FORMAT_PROMPT
+    )
+
+
+def default_tool_extractor(content: str) -> Union[str, Tuple[str, str]]:
+    regex = re.compile(r"Action:\s*([a-zA-Z0-9_]+).*?Action Input:\s*(.*)", re.DOTALL)
+    action_match = re.search(regex, content)
+    if not action_match:
+        return content
+
+    tool_name = action_match.group(1).strip()
+    tool_input = action_match.group(2).strip().strip('"').strip("```")
+    try:
+        arguments = json.loads(tool_input)
+    except json.JSONDecodeError:
+        return content
+
+    return tool_name, json.dumps(arguments, ensure_ascii=False)
+
+
+@dataclass
+class Formatter(ABC):
+    slots: SLOTS = field(default_factory=list)
+    tool_format: Optional[Literal["default"]] = None
+
+    @abstractmethod
+    def apply(self, **kwargs) -> SLOTS: ...
+
+    def extract(self, content: str) -> Union[str, Tuple[str, str]]:
+        raise NotImplementedError
+
+
+@dataclass
+class EmptyFormatter(Formatter):
+    def __post_init__(self):
+        has_placeholder = False
+        for slot in filter(lambda s: isinstance(s, str), self.slots):
+            if re.search(r"\{\{[a-zA-Z_][a-zA-Z0-9_]*\}\}", slot):
+                has_placeholder = True
+
+        if has_placeholder:
+            raise ValueError("Empty formatter should not contain any placeholder.")
+
+    def apply(self, **kwargs) -> SLOTS:
+        return self.slots
+
+
+@dataclass
+class StringFormatter(Formatter):
+    def __post_init__(self):
+        has_placeholder = False
+        for slot in filter(lambda s: isinstance(s, str), self.slots):
+            if re.search(r"\{\{[a-zA-Z_][a-zA-Z0-9_]*\}\}", slot):
+                has_placeholder = True
+
+        if not has_placeholder:
+            raise ValueError("A placeholder is required in the string formatter.")
+
+    def apply(self, **kwargs) -> SLOTS:
+        elements = []
+        for slot in self.slots:
+            if isinstance(slot, str):
+                for name, value in kwargs.items():
+                    if not isinstance(value, str):
+                        raise RuntimeError("Expected a string, got {}".format(value))
+
+                    slot = slot.replace("{{" + name + "}}", value, 1)
+                elements.append(slot)
+            elif isinstance(slot, (dict, set)):
+                elements.append(slot)
+            else:
+                raise RuntimeError("Input must be string, set[str] or dict[str, str], got {}".format(type(slot)))
+
+        return elements
+
+
+@dataclass
+class FunctionFormatter(Formatter):
+    def __post_init__(self):
+        has_name, has_args = False, False
+        for slot in filter(lambda s: isinstance(s, str), self.slots):
+            if "{{name}}" in slot:
+                has_name = True
+            if "{{arguments}}" in slot:
+                has_args = True
+
+        if not has_name or not has_args:
+            raise ValueError("Name and arguments placeholders are required in the function formatter.")
+
+    def apply(self, **kwargs) -> SLOTS:
+        content = kwargs.pop("content")
+        try:
+            function = json.loads(content)
+            name = function["name"]
+            arguments = json.dumps(function["arguments"], ensure_ascii=False)
+        except Exception:
+            name, arguments = "", ""
+
+        elements = []
+        for slot in self.slots:
+            if isinstance(slot, str):
+                slot = slot.replace("{{name}}", name).replace("{{arguments}}", arguments)
+                elements.append(slot)
+            elif isinstance(slot, (dict, set)):
+                elements.append(slot)
+            else:
+                raise RuntimeError("Input must be string, set[str] or dict[str, str], got {}".format(type(slot)))
+
+        return elements
+
+
+@dataclass
+class ToolFormatter(Formatter):
+    def __post_init__(self):
+        if self.tool_format is None:
+            raise ValueError("Tool format was not found.")
+
+    def apply(self, **kwargs) -> SLOTS:
+        content = kwargs.pop("content")
+        try:
+            tools = json.loads(content)
+            if not len(tools):
+                return [""]
+
+            if self.tool_format == "default":
+                return [default_tool_formatter(tools)]
+            else:
+                raise NotImplementedError
+        except Exception:
+            return [""]
+
+    def extract(self, content: str) -> Union[str, Tuple[str, str]]:
+        if self.tool_format == "default":
+            return default_tool_extractor(content)
+        else:
+            raise NotImplementedError

finetune/src/llmfactory/data/loader.py

+import inspect
+import os
+from typing import TYPE_CHECKING, Literal, Optional, Union
+
+from datasets import load_dataset, load_from_disk
+
+from ..extras.constants import FILEEXT2TYPE
+from ..extras.logging import get_logger
+from ..extras.misc import has_tokenized_data
+from .aligner import align_dataset
+from .parser import get_dataset_list
+from .preprocess import get_preprocess_and_print_func
+from .template import get_template_and_fix_tokenizer
+from .utils import merge_dataset
+
+
+if TYPE_CHECKING:
+    from datasets import Dataset, IterableDataset
+    from transformers import ProcessorMixin, Seq2SeqTrainingArguments
+    from transformers.tokenization_utils import PreTrainedTokenizer
+
+    from ..hparams import DataArguments, ModelArguments
+    from .parser import DatasetAttr
+
+
+logger = get_logger(__name__)
+
+
+def load_single_dataset(
+    dataset_attr: "DatasetAttr",
+    model_args: "ModelArguments",
+    data_args: "DataArguments",
+) -> Union["Dataset", "IterableDataset"]:
+    logger.info("Loading dataset {}...".format(dataset_attr))
+    data_path, data_name, data_dir, data_files = None, None, None, None
+    if dataset_attr.load_from in ["hf_hub", "ms_hub"]:
+        data_path = dataset_attr.dataset_name
+        data_name = dataset_attr.subset
+        data_dir = dataset_attr.folder
+
+    elif dataset_attr.load_from == "script":
+        data_path = os.path.join(data_args.dataset_dir, dataset_attr.dataset_name)
+        data_name = dataset_attr.subset
+        data_dir = dataset_attr.folder
+
+    elif dataset_attr.load_from == "file":
+        data_files = []
+        local_path = os.path.join(data_args.dataset_dir, dataset_attr.dataset_name)
+        if os.path.isdir(local_path):  # is directory
+            for file_name in os.listdir(local_path):
+                data_files.append(os.path.join(local_path, file_name))
+                if data_path is None:
+                    data_path = FILEEXT2TYPE.get(file_name.split(".")[-1], None)
+                elif data_path != FILEEXT2TYPE.get(file_name.split(".")[-1], None):
+                    raise ValueError("File types should be identical.")
+        elif os.path.isfile(local_path):  # is file
+            data_files.append(local_path)
+            data_path = FILEEXT2TYPE.get(local_path.split(".")[-1], None)
+        else:
+            raise ValueError("File {} not found.".format(local_path))
+
+        if data_path is None:
+            raise ValueError("File extension must be txt, csv, json or jsonl.")
+    else:
+        raise NotImplementedError
+
+    if dataset_attr.load_from == "ms_hub":
+        try:
+            from modelscope import MsDataset
+            from modelscope.utils.config_ds import MS_DATASETS_CACHE
+
+            cache_dir = model_args.cache_dir or MS_DATASETS_CACHE
+            dataset = MsDataset.load(
+                dataset_name=data_path,
+                subset_name=data_name,
+                data_dir=data_dir,
+                data_files=data_files,
+                split=data_args.split,
+                cache_dir=cache_dir,
+                token=model_args.ms_hub_token,
+                use_streaming=(data_args.streaming and (dataset_attr.load_from != "file")),
+            )
+            if isinstance(dataset, MsDataset):
+                dataset = dataset.to_hf_dataset()
+        except ImportError:
+            raise ImportError("Please install modelscope via `pip install modelscope -U`")
+    else:
+        if "trust_remote_code" in inspect.signature(load_dataset).parameters:  # for datasets==2.16.0
+            kwargs = {"trust_remote_code": True}
+        else:
+            kwargs = {}
+
+        dataset = load_dataset(
+            path=data_path,
+            name=data_name,
+            data_dir=data_dir,
+            data_files=data_files,
+            split=data_args.split,
+            cache_dir=model_args.cache_dir,
+            token=model_args.hf_hub_token,
+            streaming=(data_args.streaming and (dataset_attr.load_from != "file")),
+            **kwargs,
+        )
+
+    if data_args.streaming and (dataset_attr.load_from == "file"):  # faster than specifying streaming=True
+        dataset = dataset.to_iterable_dataset()  # TODO: add num shards parameter
+
+    if data_args.max_samples is not None:  # truncate dataset
+        num_samples = min(data_args.max_samples, len(dataset))
+        dataset = dataset.select(range(num_samples))
+
+    return align_dataset(dataset, dataset_attr, data_args)
+
+
+def get_dataset(
+    model_args: "ModelArguments",
+    data_args: "DataArguments",
+    training_args: "Seq2SeqTrainingArguments",
+    stage: Literal["pt", "sft", "rm", "kto"],
+    tokenizer: "PreTrainedTokenizer",
+    processor: Optional["ProcessorMixin"] = None,
+) -> Union["Dataset", "IterableDataset"]:
+    template = get_template_and_fix_tokenizer(tokenizer, data_args.template)
+    if data_args.train_on_prompt and template.efficient_eos:
+        raise ValueError("Current template does not support `train_on_prompt`.")
+
+    # Load tokenized dataset
+    if data_args.tokenized_path is not None:
+        if has_tokenized_data(data_args.tokenized_path):
+            logger.warning("Loading dataset from disk will ignore other data arguments.")
+            dataset = load_from_disk(data_args.tokenized_path)
+            logger.info("Loaded tokenized dataset from {}.".format(data_args.tokenized_path))
+            if data_args.streaming:
+                dataset = dataset.to_iterable_dataset()
+            return dataset
+
+        if data_args.streaming:
+            raise ValueError("Turn off `streaming` when saving dataset to disk.")
+
+    with training_args.main_process_first(desc="load dataset"):
+        all_datasets = []
+        for dataset_attr in get_dataset_list(data_args):
+            if (stage == "rm" and dataset_attr.ranking is False) or (stage != "rm" and dataset_attr.ranking is True):
+                raise ValueError("The dataset is not applicable in the current training stage.")
+
+            all_datasets.append(load_single_dataset(dataset_attr, model_args, data_args))
+        dataset = merge_dataset(all_datasets, data_args, training_args)
+
+    with training_args.main_process_first(desc="pre-process dataset"):
+        preprocess_func, print_function = get_preprocess_and_print_func(
+            data_args, training_args, stage, template, tokenizer, processor
+        )
+        column_names = list(next(iter(dataset)).keys())
+        kwargs = {}
+        if not data_args.streaming:
+            kwargs = dict(
+                num_proc=data_args.preprocessing_num_workers,
+                load_from_cache_file=(not data_args.overwrite_cache),
+                desc="Running tokenizer on dataset",
+            )
+
+        dataset = dataset.map(preprocess_func, batched=True, remove_columns=column_names, **kwargs)
+
+        if data_args.tokenized_path is not None:
+            if training_args.should_save:
+                dataset.save_to_disk(data_args.tokenized_path)
+                logger.info("Tokenized dataset saved at {}.".format(data_args.tokenized_path))
+                logger.info("Please restart the training with `--tokenized_path {}`.".format(data_args.tokenized_path))
+
+            exit(0)
+
+        if training_args.should_log:
+            try:
+                print_function(next(iter(dataset)))
+            except StopIteration:
+                raise RuntimeError("Cannot find valid samples, check `data/README.md` for the data format.")
+
+        return dataset

finetune/src/llmfactory/data/parser.py

+import json
+import os
+from dataclasses import dataclass
+from typing import TYPE_CHECKING, Any, Dict, List, Literal, Optional
+
+from ..extras.constants import DATA_CONFIG
+from ..extras.misc import use_modelscope
+
+
+if TYPE_CHECKING:
+    from ..hparams import DataArguments
+
+
+@dataclass
+class DatasetAttr:
+    r"""
+    Dataset attributes.
+    """
+
+    """ basic configs """
+    load_from: Literal["hf_hub", "ms_hub", "script", "file"]
+    dataset_name: str
+    """ extra configs """
+    subset: Optional[str] = None
+    folder: Optional[str] = None
+    ranking: bool = False
+    formatting: Literal["alpaca", "sharegpt"] = "alpaca"
+    """ common columns """
+    system: Optional[str] = None
+    tools: Optional[str] = None
+    images: Optional[str] = None
+    """ rlhf columns """
+    chosen: Optional[str] = None
+    rejected: Optional[str] = None
+    kto_tag: Optional[str] = None
+    """ alpaca columns """
+    prompt: Optional[str] = "instruction"
+    query: Optional[str] = "input"
+    response: Optional[str] = "output"
+    history: Optional[str] = None
+    """ sharegpt columns """
+    messages: Optional[str] = "conversations"
+    """ sharegpt tags """
+    role_tag: Optional[str] = "from"
+    content_tag: Optional[str] = "value"
+    user_tag: Optional[str] = "human"
+    assistant_tag: Optional[str] = "gpt"
+    observation_tag: Optional[str] = "observation"
+    function_tag: Optional[str] = "function_call"
+    system_tag: Optional[str] = "system"
+
+    def __repr__(self) -> str:
+        return self.dataset_name
+
+    def set_attr(self, key: str, obj: Dict[str, Any], default: Optional[Any] = None) -> None:
+        setattr(self, key, obj.get(key, default))
+
+
+def get_dataset_list(data_args: "DataArguments") -> List["DatasetAttr"]:
+    if data_args.dataset is not None:
+        dataset_names = [ds.strip() for ds in data_args.dataset.split(",")]
+    else:
+        dataset_names = []
+
+    if data_args.dataset_dir == "ONLINE":
+        dataset_info = None
+    else:
+        try:
+            with open(os.path.join(data_args.dataset_dir, DATA_CONFIG), "r") as f:
+                dataset_info = json.load(f)
+        except Exception as err:
+            if len(dataset_names) != 0:
+                raise ValueError(
+                    "Cannot open {} due to {}.".format(os.path.join(data_args.dataset_dir, DATA_CONFIG), str(err))
+                )
+            dataset_info = None
+
+    if data_args.interleave_probs is not None:
+        data_args.interleave_probs = [float(prob.strip()) for prob in data_args.interleave_probs.split(",")]
+
+    dataset_list: List[DatasetAttr] = []
+    for name in dataset_names:
+        if dataset_info is None:
+            load_from = "ms_hub" if use_modelscope() else "hf_hub"
+            dataset_attr = DatasetAttr(load_from, dataset_name=name)
+            dataset_list.append(dataset_attr)
+            continue
+
+        if name not in dataset_info:
+            raise ValueError("Undefined dataset {} in {}.".format(name, DATA_CONFIG))
+
+        has_hf_url = "hf_hub_url" in dataset_info[name]
+        has_ms_url = "ms_hub_url" in dataset_info[name]
+
+        if has_hf_url or has_ms_url:
+            if (use_modelscope() and has_ms_url) or (not has_hf_url):
+                dataset_attr = DatasetAttr("ms_hub", dataset_name=dataset_info[name]["ms_hub_url"])
+            else:
+                dataset_attr = DatasetAttr("hf_hub", dataset_name=dataset_info[name]["hf_hub_url"])
+        elif "script_url" in dataset_info[name]:
+            dataset_attr = DatasetAttr("script", dataset_name=dataset_info[name]["script_url"])
+        else:
+            dataset_attr = DatasetAttr("file", dataset_name=dataset_info[name]["file_name"])
+
+        dataset_attr.set_attr("subset", dataset_info[name])
+        dataset_attr.set_attr("folder", dataset_info[name])
+        dataset_attr.set_attr("ranking", dataset_info[name], default=False)
+        dataset_attr.set_attr("formatting", dataset_info[name], default="alpaca")
+
+        if "columns" in dataset_info[name]:
+            column_names = ["system", "tools", "images", "chosen", "rejected", "kto_tag"]
+            if dataset_attr.formatting == "alpaca":
+                column_names.extend(["prompt", "query", "response", "history"])
+            else:
+                column_names.extend(["messages"])
+
+            for column_name in column_names:
+                dataset_attr.set_attr(column_name, dataset_info[name]["columns"])
+
+        if dataset_attr.formatting == "sharegpt" and "tags" in dataset_info[name]:
+            tag_names = (
+                "role_tag",
+                "content_tag",
+                "user_tag",
+                "assistant_tag",
+                "observation_tag",
+                "function_tag",
+                "system_tag",
+            )
+            for tag in tag_names:
+                dataset_attr.set_attr(tag, dataset_info[name]["tags"])
+
+        dataset_list.append(dataset_attr)
+
+    return dataset_list

finetune/src/llmfactory/data/utils.py

+from enum import Enum, unique
+from typing import TYPE_CHECKING, Dict, List, Tuple, Union
+
+from datasets import concatenate_datasets, interleave_datasets
+
+from ..extras.logging import get_logger
+
+
+if TYPE_CHECKING:
+    from datasets import Dataset, IterableDataset
+    from transformers import Seq2SeqTrainingArguments
+
+    from ..hparams import DataArguments
+
+
+logger = get_logger(__name__)
+
+
+@unique
+class Role(str, Enum):
+    USER = "user"
+    ASSISTANT = "assistant"
+    SYSTEM = "system"
+    FUNCTION = "function"
+    OBSERVATION = "observation"
+
+
+def infer_max_len(source_len: int, target_len: int, max_len: int, reserved_label_len: int) -> Tuple[int, int]:
+    max_target_len = int(max_len * (target_len / (source_len + target_len)))
+    max_target_len = max(max_target_len, reserved_label_len)
+    max_source_len = max_len - min(max_target_len, target_len)
+    return max_source_len, max_target_len
+
+
+def merge_dataset(
+    all_datasets: List[Union["Dataset", "IterableDataset"]],
+    data_args: "DataArguments",
+    training_args: "Seq2SeqTrainingArguments",
+) -> Union["Dataset", "IterableDataset"]:
+    if len(all_datasets) == 1:
+        return all_datasets[0]
+    elif data_args.mix_strategy == "concat":
+        if data_args.streaming:
+            logger.warning("The samples between different datasets will not be mixed in streaming mode.")
+        return concatenate_datasets(all_datasets)
+    elif data_args.mix_strategy.startswith("interleave"):
+        if not data_args.streaming:
+            logger.warning("We recommend using `mix_strategy=concat` in non-streaming mode.")
+        return interleave_datasets(
+            datasets=all_datasets,
+            probabilities=data_args.interleave_probs,
+            seed=training_args.seed,
+            stopping_strategy="first_exhausted" if data_args.mix_strategy.endswith("under") else "all_exhausted",
+        )
+    else:
+        raise ValueError("Unknown mixing strategy.")
+
+
+def split_dataset(
+    dataset: Union["Dataset", "IterableDataset"], data_args: "DataArguments", training_args: "Seq2SeqTrainingArguments"
+) -> Dict[str, "Dataset"]:
+    if training_args.do_train:
+        if data_args.val_size > 1e-6:  # Split the dataset
+            if data_args.streaming:
+                dataset = dataset.shuffle(buffer_size=data_args.buffer_size, seed=training_args.seed)
+                val_set = dataset.take(int(data_args.val_size))
+                train_set = dataset.skip(int(data_args.val_size))
+                return {"train_dataset": train_set, "eval_dataset": val_set}
+            else:
+                val_size = int(data_args.val_size) if data_args.val_size > 1 else data_args.val_size
+                dataset = dataset.train_test_split(test_size=val_size, seed=training_args.seed)
+                return {"train_dataset": dataset["train"], "eval_dataset": dataset["test"]}
+        else:
+            if data_args.streaming:
+                dataset = dataset.shuffle(buffer_size=data_args.buffer_size, seed=training_args.seed)
+            return {"train_dataset": dataset}
+    else:  # do_eval or do_predict
+        return {"eval_dataset": dataset}

finetune/src/llmfactory/eval/template.py

+from dataclasses import dataclass
+from typing import Dict, List, Sequence, Tuple
+
+from ..data import Role
+from ..extras.constants import CHOICES
+
+
+@dataclass
+class EvalTemplate:
+    system: str
+    choice: str
+    answer: str
+    prefix: str
+
+    def _parse_example(self, example: Dict[str, str]) -> Tuple[str, str]:
+        r"""
+        input: a dict with keys {"question", "A", "B", "C", "D", "answer"}
+        output: a tuple of (prompt, response)
+        """
+        candidates = [self.choice.format(choice=ch, content=example[ch]) for ch in CHOICES if ch in example]
+        return "".join([example["question"]] + candidates + [self.answer]), example["answer"]
+
+    def format_example(
+        self, target_data: Dict[str, str], support_set: Sequence[Dict[str, str]], subject_name: str
+    ) -> List[Dict[str, str]]:
+        r"""
+        Converts dataset examples to messages.
+        """
+        messages = []
+        for k in range(len(support_set)):
+            prompt, response = self._parse_example(support_set[k])
+            messages.append({"role": Role.USER.value, "content": prompt})
+            messages.append({"role": Role.ASSISTANT.value, "content": response})
+
+        prompt, response = self._parse_example(target_data)
+        messages.append({"role": Role.USER.value, "content": prompt})
+        messages.append({"role": Role.ASSISTANT.value, "content": response})
+        messages[0]["content"] = self.system.format(subject=subject_name) + messages[0]["content"]
+        return messages
+
+
+eval_templates: Dict[str, "EvalTemplate"] = {}
+
+
+def _register_eval_template(name: str, system: str, choice: str, answer: str, prefix: str) -> None:
+    eval_templates[name] = EvalTemplate(system=system, choice=choice, answer=answer, prefix=prefix)
+
+
+def get_eval_template(name: str) -> "EvalTemplate":
+    eval_template = eval_templates.get(name, None)
+    assert eval_template is not None, "Template {} does not exist.".format(name)
+    return eval_template
+
+
+_register_eval_template(
+    name="en",
+    system="The following are multiple choice questions (with answers) about {subject}.\n\n",
+    choice="\n{choice}. {content}",
+    answer="\nAnswer: ",
+    prefix=" ",
+)
+
+
+_register_eval_template(
+    name="zh",
+    system="以下是中国关于{subject}考试的单项选择题，请选出其中的正确答案。\n\n",
+    choice="\n{choice}. {content}",
+    answer="\n答案：",
+    prefix=" ",
+)