Qwen源码解析：finetune.py

# This code is based on the revised code from fastchat based on tatsu-lab/stanford_alpaca.
 
 
from dataclasses import dataclass, field
import json
import math
import logging
import os
from typing import Dict, Optional, List
import torch
from torch.utils.data import Dataset
from deepspeed import zero
from deepspeed.runtime.zero.partition_parameters import ZeroParamStatus
import transformers
from transformers import Trainer, GPTQConfig, deepspeed
from transformers.trainer_pt_utils import LabelSmoother
from peft import LoraConfig, get_peft_model, prepare_model_for_kbit_training
 
 
IGNORE_TOKEN_ID = LabelSmoother.ignore_index
 
 
@dataclass
class ModelArguments:
    model_name_or_path: Optional[str] = field(default="Qwen/Qwen-7B")
 
 
@dataclass
class DataArguments:
    data_path: str = field(
        default=None, metadata={"help": "Path to the training data."}
    )
    eval_data_path: str = field(
        default=None, metadata={"help": "Path to the evaluation data."}
    )
    lazy_preprocess: bool = False
 
 
@dataclass
class TrainingArguments(transformers.TrainingArguments):
    cache_dir: Optional[str] = field(default=None)
    optim: str = field(default="adamw_torch")
    model_max_length: int = field(
        default=8192,
        metadata={
            "help": "Maximum sequence length. Sequences will be right padded (and possibly truncated)."
        },
    )
    use_lora: bool = False
 
 
@dataclass
class LoraArguments:
    lora_r: int = 64
    lora_alpha: int = 16
    lora_dropout: float = 0.05
    lora_target_modules: List[str] = field(
        default_factory=lambda: ["c_attn", "c_proj", "w1", "w2"]
    )
    lora_weight_path: str = ""
    lora_bias: str = "none"
    q_lora: bool = False
 
 
def maybe_zero_3(param):
    if hasattr(param, "ds_id"):
        assert param.ds_status == ZeroParamStatus.NOT_AVAILABLE
        with zero.GatheredParameters([param]):
            param = param.data.detach().cpu().clone()
    else:
        param = param.detach().cpu().clone()
    return param
 
 
# Borrowed from peft.utils.get_peft_model_state_dict
def get_peft_state_maybe_zero_3(named_params, bias):
    if bias == "none":
        to_return = {k: t for k, t in named_params if "lora_" in k}
    elif bias == "all":
        to_return = {k: t for k, t in named_params if "lora_" in k or "bias" in k}
    elif bias == "lora_only":
        to_return = {}
        maybe_lora_bias = {}
        lora_bias_names = set()
        for k, t in named_params:
            if "lora_" in k:
                to_return[k] = t
                bias_name = k.split("lora_")[0] + "bias"
                lora_bias_names.add(bias_name)
            elif "bias" in k:
                maybe_lora_bias[k] = t
        for k, t in maybe_lora_bias:
            if bias_name in lora_bias_names:
                to_return[bias_name] = t
    else:
        raise NotImplementedError
    to_return = {k: maybe_zero_3(v) for k, v in to_return.items()}
    return to_return
 
local_rank = None
 
def rank0_print(*args):
    if local_rank == 0:
        print(*args)
 
 
def safe_save_model_for_hf_trainer(trainer: transformers.Trainer, output_dir: str, bias="none"):
    """Collects the state dict and dump to disk."""
    # check if zero3 mode enabled
    if deepspeed.is_deepspeed_zero3_enabled():
        state_dict = trainer.model_wrapped._zero3_consolidated_16bit_state_dict()
    else:
        if trainer.args.use_lora:
            state_dict = get_peft_state_maybe_zero_3(
                trainer.model.named_parameters(), bias
            )
        else:
            state_dict = trainer.model.state_dict()
    if trainer.args.should_save and trainer.args.local_rank == 0:
        trainer._save(output_dir, state_dict=state_dict)
 
 
def preprocess(
    sources,
    tokenizer: transformers.PreTrainedTokenizer,
    max_len: int,
    system_message: str = "You are a helpful assistant."
) -> Dict:
    roles = {"user": "<|im_start|>user", "assistant": "<|im_start|>assistant"}
 
    im_start = tokenizer.im_start_id
    im_end = tokenizer.im_end_id
    nl_tokens = tokenizer('\n').input_ids
    _system = tokenizer('system').input_ids + nl_tokens
    _user = tokenizer('user').input_ids + nl_tokens
    _assistant = tokenizer('assistant').input_ids + nl_tokens
 
    # Apply prompt templates
    input_ids, targets = [], []
    for i, source in enumerate(sources):
        if roles[source[0]["from"]] != roles["user"]:
            source = source[1:]
 
        input_id, target = [], []
        system = [im_start] + _system + tokenizer(system_message).input_ids + [im_end] + nl_tokens
        input_id += system
        target += [im_start] + [IGNORE_TOKEN_ID] * (len(system)-3) + [im_end] + nl_tokens
        assert len(input_id) == len(target)
        for j, sentence in enumerate(source):
            role = roles[sentence["from"]]
            _input_id = tokenizer(role).input_ids + nl_tokens + \
                tokenizer(sentence["value"]).input_ids + [im_end] + nl_tokens
            input_id += _input_id
            if role == '<|im_start|>user':
                _target = [im_start] + [IGNORE_TOKEN_ID] * (len(_input_id)-3) + [im_end] + nl_tokens
            elif role == '<|im_start|>assistant':
                _target = [im_start] + [IGNORE_TOKEN_ID] * len(tokenizer(role).input_ids) + \
                    _input_id[len(tokenizer(role).input_ids)+1:-2] + [im_end] + nl_tokens
            else:
                raise NotImplementedError
            target += _target
        assert len(input_id) == len(target)
        input_id += [tokenizer.pad_token_id] * (max_len - len(input_id))
        target += [IGNORE_TOKEN_ID] * (max_len - len(target))
        input_ids.append(input_id[:max_len])
        targets.append(target[:max_len])
    input_ids = torch.tensor(input_ids, dtype=torch.int)
    targets = torch.tensor(targets, dtype=torch.int)
 
    return dict(
        input_ids=input_ids,
        labels=targets,
        attention_mask=input_ids.ne(tokenizer.pad_token_id),
    )
 
 
class SupervisedDataset(Dataset):
    """Dataset for supervised fine-tuning."""
 
    def __init__(self, raw_data, tokenizer: transformers.PreTrainedTokenizer, max_len: int):
        super(SupervisedDataset, self).__init__()
 
        rank0_print("Formatting inputs...")
        sources = [example["conversations"] for example in raw_data]
        data_dict = preprocess(sources, tokenizer, max_len)
 
        self.input_ids = data_dict["input_ids"]
        self.labels = data_dict["labels"]
        self.attention_mask = data_dict["attention_mask"]
 
    def __len__(self):
        return len(self.input_ids)
 
    def __getitem__(self, i) -> Dict[str, torch.Tensor]:
        return dict(
            input_ids=self.input_ids[i],
            labels=self.labels[i],
            attention_mask=self.attention_mask[i],
        )

Python代码，涉及deepspeed、transformers、peft等库，用于定义数据与模型相关的参数和配置。下面我会逐行解释这段代码的含义：

pythonCopy code

from deepspeed import zero

• 从deepspeed库中导入zero模块。DeepSpeed是一个深度学习优化库，zero是它的一个特性，可以减少模型的内存占用。

pythonCopy code

from deepspeed.runtime.zero.partition_parameters import ZeroParamStatus

• 从deepspeed的zero模块中导入ZeroParamStatus。它与模型参数的分区有关，通常与zero优化有关。

pythonCopy code

import transformers

• 导入transformers库。transformers是一个提供大量预训练模型的库，例如BERT、GPT-2等。

pythonCopy code

from transformers import Trainer, GPTQConfig, deepspeed

• 从transformers库中导入Trainer、GPTQConfig和deepspeed。Trainer是用于训练模型的类，GPTQConfig可能是针对某种GPT模型的配置，而deepspeed是与DeepSpeed集成相关的模块。

pythonCopy code

from transformers.trainer_pt_utils import LabelSmoother

• 从transformers的trainer_pt_utils模块中导入LabelSmoother。LabelSmoother是一个用于标签平滑的工具，可以提高模型的泛化能力。

pythonCopy code

from peft import LoraConfig, get_peft_model, prepare_model_for_kbit_training

• 从peft库中导入LoraConfig、get_peft_model和prepare_model_for_kbit_training。不过，到目前为止（2022年1月）我不熟悉peft这个库，所以具体的功能需要查阅相关文档。

pythonCopy code

IGNORE_TOKEN_ID = LabelSmoother.ignore_index

• 定义一个常量IGNORE_TOKEN_ID，其值等于LabelSmoother的ignore_index属性。这通常用于指定在计算损失时应该忽略的token ID。

pythonCopy code

@dataclass class ModelArguments: model_name_or_path: Optional[str] = field(default="Qwen/Qwen-7B")

• 使用Python的dataclass定义一个名为ModelArguments的类，它有一个属性model_name_or_path，默认值为"Qwen/Qwen-7B"。

pythonCopy code

@dataclass class DataArguments: data_path: str = field( default=None, metadata={"help": "Path to the training data."} ) eval_data_path: str = field( default=None, metadata={"help": "Path to the evaluation data."} ) lazy_preprocess: bool = False

• 使用dataclass定义一个名为DataArguments的类，用于指定数据相关的参数，如训练数据路径、评估数据路径和是否使用懒加载预处理。

pythonCopy code

@dataclass class TrainingArguments(transformers.TrainingArguments): cache_dir: Optional[str] = field(default=None) optim: str = field(default="adamw_torch") model_max_length: int = field( default=8192, metadata={ "help": "Maximum sequence length. Sequences will be right padded (and possibly truncated)." }, ) use_lora: bool = False

• 使用dataclass定义一个名为TrainingArguments的类，它继承了transformers的TrainingArguments类。这个类定义了训练相关的参数，如缓存目录、优化器、模型的最大序列长度和是否使用Lora。

这段代码主要定义了与模型、数据和训练相关的参数和配置类。希望这个解释对您有所帮助！