LLAVA代码阅读：train.py

源码路径:https://github.com/haotian-liu/LLaVA/llava/model/train/train.py

声明：此博客为个人学习笔记，可能有理解错误的地方，敬请指正

函数定义

def train(attn_implementation=None):
    global local_rank

attn_implementation:注意力机制接口，此参数决定模型使用何种注意力机制，此参数会在后面模型定义处使用

local_rank:在分布式训练中使用，表示当前训练实例（进程）的本地排名或编号。在分布式训练时，local_rank 用于指定每个进程的任务（比如在多卡训练时，每张GPU会有一个 local_rank）

参数解析

parser = transformers.HfArgumentParser(
        (ModelArguments, DataArguments, TrainingArguments))
    model_args, data_args, training_args = parser.parse_args_into_dataclasses()
    local_rank = training_args.local_rank
    compute_dtype = (torch.float16 if training_args.fp16 else (torch.bfloat16 if training_args.bf16 else torch.float32))

1.使用HuggingFace提供的专门解析命令行参数的类，并返回类实例

2.根据解析后的参数给local_rank和compute_dtype赋值

模型加载与量化参数配置

    bnb_model_from_pretrained_args = {}
    if training_args.bits in [4, 8]:
        from transformers import BitsAndBytesConfig
        bnb_model_from_pretrained_args.update(dict(
            device_map={"": training_args.device},
            load_in_4bit=training_args.bits == 4,
            load_in_8bit=training_args.bits == 8,
            quantization_config=BitsAndBytesConfig(
                load_in_4bit=training_args.bits == 4,
                load_in_8bit=training_args.bits == 8,
                llm_int8_skip_modules=["mm_projector"],
                llm_int8_threshold=6.0,
                llm_int8_has_fp16_weight=False,
                bnb_4bit_compute_dtype=compute_dtype,
                bnb_4bit_use_double_quant=training_args.double_quant,
                bnb_4bit_quant_type=training_args.quant_type # {'fp4', 'nf4'}
            )
        ))

此处根据training_args中的参数构造了字典bnb_model_from_pretrained_args，配置了与bits_and_bytes相关的模型加载参数，特别是量化时加载模型的方式，为的是减少显存占用

定义模型：获得model

    if model_args.vision_tower is not None:
        if 'mpt' in model_args.model_name_or_path:
            config = transformers.AutoConfig.from_pretrained(model_args.model_name_or_path, trust_remote_code=True)
            config.attn_config['attn_impl'] = training_args.mpt_attn_impl
            model = LlavaMptForCausalLM.from_pretrained(
                model_args.model_name_or_path,
                config=config,
                cache_dir=training_args.cache_dir,
                **bnb_model_from_pretrained_args
            )
        else:
            model = LlavaLlamaForCausalLM.from_pretrained(
                model_args.model_name_or_path,
                cache_dir=training_args.cache_dir,
                attn_implementation=attn_implementation,
                torch_dtype=(torch.bfloat16 if training_args.bf16 else None),
                **bnb_model_from_pretrained_args
            )
    else:
        model = transformers.LlamaForCausalLM.from_pretrained(
            model_args.model_name_or_path,
            cache_dir=training_args.cache_dir,
            attn_implementation=attn_implementation,
            torch_dtype=(torch.bfloat16 if training_args.bf16 else None),
            **bnb_model_from_pretrained_args
        )
    model.config.use_cache = False

根据是否有视觉塔来定义不同的模型。可见LLAVA工程train函数的复用性较强，可支持多种模型的训练

冻结骨干网络

if model_args.freeze_backbone:
    model.model.requires_grad_(False)

冻结骨干网络，只对新加入的层进行训练更新，有效防止过拟合且减小计算量。很好理解：一些预训练过的东西就没有必要动它了

进一步配置量化训练

    if training_args.bits in [4, 8]:
        from peft import prepare_model_for_kbit_training
        model.config.torch_dtype=(torch.float32 if training_args.fp16 else (torch.bfloat16 if training_args.bf16 else torch.float32))
        model = prepare_model_for_kbit_training(model, use_gradient_checkpointing=training_args.gradient_checkpointing)

PEFT（Parameter Efficient Fine-Tuning）通常用于高效微调大模型  

torch_dtype:模型精度

为低精度训练做好准备，并设置梯度检查点

启动部分层的梯度计算

    if training_args.gradient_checkpointing:
        if hasattr(model, "enable_input_require_grads"):
            model.enable_input_require_grads()
        else:
            def make_inputs_require_grad(module, input, output):
                output.requires_grad_(True)
            model.get_input_embeddings().register_forward_hook(make_inputs_require_grad)

register_forward_hook :是 PyTorch 中的一个方法，它允许你在模型前向传播时插入一个钩子函数（hook）。该钩子函数会在模型执行前向传播时被调用，make_inputs_require_grad 就是在这个时候被触发的。通过这个钩子函数，每当模型的输入嵌入层（input_embeddings）被调用时，它都会确保其输出的 requires_grad 属性被设置为 True，即启用梯度计算。

LoRA配置和模型转换

LoRA 是一种用于有效微调预训练模型的技术，特别是对于大模型的参数高效调整

 if training_args.lora_enable:
        from peft import LoraConfig, get_peft_model
        lora_config = LoraConfig(
            r=training_args.lora_r,
            lora_alpha=training_args.lora_alpha,
            target_modules=find_all_linear_names(model),
            lora_dropout=training_args.lora_dropout,
            bias=training_args.lora_bias,
            task_type="CAUSAL_LM",
        )
        if training_args.bits == 16:
            if training_args.bf16:
                model.to(torch.bfloat16)
            if training_args.fp16:
                model.to(torch.float16)
        rank0_print("Adding LoRA adapters...")
        model = get_peft_model(model, lora_config)

r:LoRA适配器的秩

lora_alpha:放大因子，决定LORA的影响力

target_moudles:目标模块，这里是一些线性层

rank0_print:主进程打印提示信息

get_peft_model:对模型进行进一步处理

定义Tokenizer

    if 'mpt' in model_args.model_name_or_path:
        tokenizer = transformers.AutoTokenizer.from_pretrained(
            model_args.model_name_or_path,
            cache_dir=training_args.cache_dir,
            model_max_length=training_args.model_max_length,
            padding_side="right"
        )
    else:
        tokenizer = transformers.AutoTokenizer.from_pretrained(
            model_args.model_name_or_path,
            cache_dir=training_args.cache_dir,
            model_max_length=training_args.model_max_length,
            padding_side="right",
            use_fast=False,
        )

mpt:Multi-Modal Pretrained Transformer是一种能够处理多模态输入的 Transformer 模型

use_fast:是否使用fast tokenizer比较高效，但不支持一些情况，所以在第二种情况下使用的是标准Python tokenizer

配置Tokenizer

    if model_args.version == "v0":
        if tokenizer.pad_token is None:
            smart_tokenizer_and_embedding_resize(
                special_tokens_dict=dict(pad_token="[PAD]"),
                tokenizer=tokenizer,
                model=model,
            )
    elif model_args.version == "v0.5":
        tokenizer.pad_token = tokenizer.unk_token
    else:
        tokenizer.pad_token = tokenizer.unk_token
        if model_args.version in conversation_lib.conv_templates:
            conversation_lib.default_conversation = conversation_lib.conv_templates[model_args.version]
        else:
            conversation_lib.default_conversation = conversation_lib.conv_templates["vicuna_v1"]

vision_tower和vision_tokenizer初始化

    if model_args.vision_tower is not None:
        model.get_model().initialize_vision_modules(
            model_args=model_args,
            fsdp=training_args.fsdp
        )

        vision_tower = model.get_vision_tower()
        vision_tower.to(dtype=torch.bfloat16 if training_args.bf16 else torch.float16, device=training_args.device)

        data_args.image_processor = vision_tower.image_processor
        data_args.is_multimodal = True

        model.config.image_aspect_ratio = data_args.image_aspect_ratio
        model.config.tokenizer_padding_side = tokenizer.padding_side
        model.config.tokenizer_model_max_length = tokenizer.model_max_length

        model.config.tune_mm_mlp_adapter = training_args.tune_mm_mlp_adapter = model_args.tune_mm_mlp_adapter
        if model_args.tune_mm_mlp_adapter:
            model.requires_grad_(False)
            for p in model.get_model().mm_projector.parameters():
                p.requires_grad = True

        model.config.freeze_mm_mlp_adapter = training_args.freeze_mm_mlp_adapter
        if training_args.freeze_mm_mlp_adapter:
            for p in model.get_model().mm_projector.parameters():
                p.requires_grad = False

        if training_args.bits in [4, 8]:
            model.get_model().mm_projector.to(dtype=compute_dtype, device=training_args.device)

        model.config.mm_use_im_start_end = data_args.mm_use_im_start_end = model_args.mm_use_im_start_end
        model.config.mm_projector_lr = training_args.mm_projector_lr
        training_args.use_im_start_end = model_args.mm_use_im_start_end
        model.config.mm_use_im_patch_token = model_args.mm_use_im_patch_token
        model.initialize_vision_tokenizer(model_args, tokenizer=tokenizer)

看似一顿操作，其实就是为了配置一个vision_tower和vision_tokenier，这两个都在model里

对模型中的部分层做数据类型转换

    if training_args.bits in [4, 8]:
        from peft.tuners.lora import LoraLayer
        for name, module in model.named_modules():
            if isinstance(module, LoraLayer):
                if training_args.bf16:
                    module = module.to(torch.bfloat16)
            if 'norm' in name:
                module = module.to(torch.float32)
            if 'lm_head' in name or 'embed_tokens' in name:
                if hasattr(module, 'weight'):
                    if training_args.bf16 and module.weight.dtype == torch.float32:
                        module = module.to(torch.bfloat16)

训练并保存模型

data_module = make_supervised_data_module(tokenizer=tokenizer,
                                              data_args=data_args)
    trainer = LLaVATrainer(model=model,
                    tokenizer=tokenizer,
                    args=training_args,
                    **data_module)

    if list(pathlib.Path(training_args.output_dir).glob("checkpoint-*")):
        trainer.train(resume_from_checkpoint=True)
    else:
        trainer.train()
    trainer.save_state()

    model.config.use_cache = True

    if training_args.lora_enable:
        state_dict = get_peft_state_maybe_zero_3(
            model.named_parameters(), training_args.lora_bias
        )
        non_lora_state_dict = get_peft_state_non_lora_maybe_zero_3(
            model.named_parameters()
        )
        if training_args.local_rank == 0 or training_args.local_rank == -1:
            model.config.save_pretrained(training_args.output_dir)
            model.save_pretrained(training_args.output_dir, state_dict=state_dict)
            torch.save(non_lora_state_dict, os.path.join(training_args.output_dir, 'non_lora_trainables.bin'))
    else:
        safe_save_model_for_hf_trainer(trainer=trainer,
                                       output_dir=training_args.output_dir)

这里trainer.save_state()是保存模型当前状态，防止中断后丢失进度，和保存模型并不同

个人总结：

LoRA是一种大规模预训练模型微调的方法，日后有时间深入学习

还有那些数据类型转换的部分理解不够深入，还需要加强学习模型量化和分布式训练的内容