Create alpaca_lora.py

alpaca_lora.py

@@ -0,0 +1,343 @@
+import os
+import sys
+from typing import List
+
+import fire
+import torch
+import transformers
+from datasets import load_dataset
+from typing import List
+import json
+
+def jload(data_path:str)-> List:
+    with open(data_path,'r') as f:
+        data = json.load(f)
+    return data
+
+"""
+Unused imports:
+import torch.nn as nn
+import bitsandbytes as bnb
+"""
+
+from peft import (
+    LoraConfig,
+    get_peft_model,
+    PeftModel,
+    get_peft_model_state_dict,
+    prepare_model_for_int8_training,
+    set_peft_model_state_dict,
+)
+from transformers import LlamaForCausalLM, LlamaTokenizer,EarlyStoppingCallback
+
+
+
+
+import json
+import os.path as osp
+from typing import Union
+
+# os.environ["WANDB_DISABLED"] = "true"
+class Prompter(object):
+    __slots__ = ("template", "_verbose")
+
+    def __init__(self, template_name: str = "", verbose: bool = False):
+        self._verbose = verbose
+        if not template_name:
+            # Enforce the default here, so the constructor can be called with '' and will not break.
+            template_name = "alpaca"
+        file_name = osp.join("data/templates", f"{template_name}.json")
+        if not osp.exists(file_name):
+            raise ValueError(f"{file_name} 文件不存在")
+        with open(file_name) as fp:
+            self.template = json.load(fp)
+        if self._verbose:
+            print(
+                f"Using prompt template {file_name}: {self.template['description']}"
+            )
+
+    def generate_prompt(
+        self,
+        instruction: str,
+        input: Union[None, str] = None,
+        label: Union[None, str] = None,
+    ) -> str:
+        # returns the full prompt from instruction and optional input
+        # if a label (=response, =output) is provided, it's also appended.
+        if input:
+            res = self.template["prompt_input"].format(
+                instruction=instruction, input=input
+            )
+        else:
+            res = self.template["prompt_no_input"].format(
+                instruction=instruction
+            )
+        if label:
+            res = f"{res}{label}"
+        if self._verbose:
+            print(res)
+        return res
+
+    def get_response(self, output: str) -> str:
+        return output.split(self.template["response_split"])[1].strip()
+
+
+def train(
+    # model/data params
+    base_model: str = "",  # the only required argument
+    data_path: str = "data/alapa",
+    output_dir: str = "./lora-alpaca",
+    # training hyperparams
+    batch_size: int = 12,
+    micro_batch_size: int = 4,
+    num_epochs: int = 3,
+    learning_rate: float = 3e-4,
+    cutoff_len: int = 512,
+    val_set_size: int = 200,
+    # lora hyperparams
+    lora_r: int = 64,
+    lora_alpha: int = 128,
+    lora_dropout: float = 0.05,
+    lora_target_modules: List[str] = [
+        "q_proj",
+        "v_proj",
+    ],
+    cache_dir=None,
+    peft_path='',
+    report_to='none',
+    # llm hyperparams
+    train_on_inputs: bool = False,  # if False, masks out inputs in loss
+    add_eos_token: bool = False,
+    group_by_length: bool = False,  # faster, but produces an odd training loss curve
+    # wandb params
+    
+    resume_from_checkpoint: str = None,  # either training checkpoint or final adapter
+    prompt_template_name: str = "alpaca",  # The prompt template to use, will default to alpaca.
+):
+    if int(os.environ.get("LOCAL_RANK", 0)) == 0:
+        print(
+            f"Training Alpaca-LoRA model with params:\n"
+            f"base_model: {base_model}\n"
+            f"data_path: {data_path}\n"
+            f"output_dir: {output_dir}\n"
+            f"batch_size: {batch_size}\n"
+            f"micro_batch_size: {micro_batch_size}\n"
+            f"num_epochs: {num_epochs}\n"
+            f"learning_rate: {learning_rate}\n"
+            f"cutoff_len: {cutoff_len}\n"
+            f"val_set_size: {val_set_size}\n"
+            f"lora_r: {lora_r}\n"
+            f"cache_dir: {cache_dir}\n"
+            f"lora_alpha: {lora_alpha}\n"
+            f"lora_dropout: {lora_dropout}\n"
+            f"lora_target_modules: {lora_target_modules}\n"
+            f"train_on_inputs: {train_on_inputs}\n"
+            f"group_by_length: {group_by_length}\n"
+            f"resume_from_checkpoint: {resume_from_checkpoint or False}\n"
+            f"prompt template: {prompt_template_name}\n"
+            f"peft_path: {peft_path}\n"
+        )
+    assert (
+        base_model
+    ), "Please specify a --base_model, e.g. --base_model='huggyllama/llama-7b'"
+    gradient_accumulation_steps = batch_size // micro_batch_size
+
+    prompter = Prompter(prompt_template_name)
+
+    device_map = "auto"
+    world_size = int(os.environ.get("WORLD_SIZE", 1))
+    ddp = world_size != 1
+    if ddp:
+        device_map = {"": int(os.environ.get("LOCAL_RANK") or 0)}
+        gradient_accumulation_steps = gradient_accumulation_steps // world_size
+
+    
+
+    model = LlamaForCausalLM.from_pretrained(
+        base_model,
+        load_in_8bit=False,
+        torch_dtype=torch.float16,
+        device_map=device_map,
+        cache_dir=cache_dir,
+    )
+    
+    tokenizer = LlamaTokenizer.from_pretrained(base_model)
+    if tokenizer.pad_token_id is None:
+        tokenizer.pad_token_id = (
+            len(tokenizer) + 1  # unk. we want this to be different from the eos token
+        )
+    tokenizer.padding_side = "left"  # Allow batched inference
+    if model.get_input_embeddings().weight.size(0) != len(tokenizer):
+        
+        print("Resize model embeddings to fit tokenizer")
+        model.resize_token_embeddings(len(tokenizer))
+        
+    def tokenize(prompt, add_eos_token=True):
+        # there's probably a way to do this with the tokenizer settings
+        # but again, gotta move fast
+        result = tokenizer(
+            prompt,
+            truncation=True,
+            max_length=cutoff_len,
+            padding=False,
+            return_tensors=None,
+        )
+        if (
+            result["input_ids"][-1] != tokenizer.eos_token_id
+            and len(result["input_ids"]) < cutoff_len
+            and add_eos_token
+        ):
+            result["input_ids"].append(tokenizer.eos_token_id)
+            result["attention_mask"].append(1)
+
+        result["labels"] = result["input_ids"].copy()
+
+        return result
+
+    def generate_and_tokenize_prompt(data_point):
+        full_prompt = prompter.generate_prompt(
+            data_point["instruction"],
+            data_point["input"],
+            data_point["output"],
+        )
+        tokenized_full_prompt = tokenize(full_prompt)
+        if not train_on_inputs:
+            user_prompt = prompter.generate_prompt(
+                data_point["instruction"], data_point["input"]
+            )
+            tokenized_user_prompt = tokenize(
+                user_prompt, add_eos_token=add_eos_token
+            )
+            user_prompt_len = len(tokenized_user_prompt["input_ids"])
+
+            if add_eos_token:
+                user_prompt_len -= 1
+
+            tokenized_full_prompt["labels"] = [
+                -100
+            ] * user_prompt_len + tokenized_full_prompt["labels"][
+                user_prompt_len:
+            ]  # could be sped up, probably
+        return tokenized_full_prompt
+
+
+    # model = prepare_model_for_int8_training(model)
+
+    config = LoraConfig(
+        r=lora_r,
+        lora_alpha=lora_alpha,
+        target_modules=lora_target_modules,
+        lora_dropout=lora_dropout,
+        bias="none",
+        task_type="CAUSAL_LM",
+    )
+    model = get_peft_model(model, config)
+    
+    
+    if data_path.endswith(".json") or data_path.endswith(".jsonl"):
+        data = load_dataset("json", data_files=data_path)
+        # data = jload(data_path)
+    else:
+        data = load_dataset(data_path)
+    
+    if resume_from_checkpoint:
+        # Check the available weights and load them
+        checkpoint_name = os.path.join(
+            resume_from_checkpoint, "pytorch_model.bin"
+        )  # Full checkpoint
+        if not os.path.exists(checkpoint_name):
+            checkpoint_name = os.path.join(
+                resume_from_checkpoint, "adapter_model.bin"
+            )  # only LoRA model - LoRA config above has to fit
+            resume_from_checkpoint = (
+                False  # So the trainer won't try loading its state
+            )
+        # The two files above have a different name depending on how they were saved, but are actually the same.
+        if os.path.exists(checkpoint_name):
+            print(f"Restarting from {checkpoint_name}")
+            adapters_weights = torch.load(checkpoint_name)
+            set_peft_model_state_dict(model, adapters_weights)
+        else:
+            print(f"Checkpoint {checkpoint_name} not found")
+    
+    if peft_path: 
+        adapters_weights = torch.load(f"{peft_path}/adapter_model.bin")
+        set_peft_model_state_dict(model, adapters_weights)
+    model.print_trainable_parameters()  # Be more transparent about the % of trainable params.
+
+    if val_set_size > 0:
+        train_val = data["train"].train_test_split(
+            test_size=val_set_size, shuffle=True, seed=42
+        )
+        train_data = (
+            train_val["train"].shuffle().map(generate_and_tokenize_prompt)
+        )
+        val_data = (
+            train_val["test"].shuffle().map(generate_and_tokenize_prompt)
+        )
+    else:
+        train_data = data["train"].shuffle().map(generate_and_tokenize_prompt)
+        val_data = None
+
+    if not ddp and torch.cuda.device_count() > 1:
+        # keeps Trainer from trying its own DataParallelism when more than 1 gpu is available
+        model.is_parallelizable = True
+        model.model_parallel = True
+
+    trainer = transformers.Trainer(
+        model=model,
+        train_dataset=train_data,
+        eval_dataset=val_data,
+        args=transformers.TrainingArguments(
+            per_device_train_batch_size=micro_batch_size,
+            gradient_accumulation_steps=gradient_accumulation_steps,
+            warmup_steps=10,
+            num_train_epochs=num_epochs,
+            learning_rate=learning_rate,
+            fp16=True,
+            sharded_ddp=True,
+            logging_steps=200,
+            optim="adamw_torch",
+            evaluation_strategy="steps" if val_set_size > 0 else "no",
+            save_strategy="steps",
+            eval_steps=200 if val_set_size > 0 else None,
+            save_steps=200,
+            logging_strategy='steps',
+            output_dir=output_dir,
+            save_total_limit=5,
+            load_best_model_at_end=True if val_set_size > 0 else False,
+            ddp_find_unused_parameters=False if ddp else None,
+            group_by_length=group_by_length,
+            report_to=report_to,
+           
+
+        ),
+        data_collator=transformers.DataCollatorForSeq2Seq(
+            tokenizer, pad_to_multiple_of=8, return_tensors="pt", padding=True
+        ),
+       
+    )
+    model.config.use_cache = False
+
+    old_state_dict = model.state_dict
+    model.state_dict = (
+        lambda self, *_, **__: get_peft_model_state_dict(
+            self, old_state_dict()
+        )
+    ).__get__(model, type(model))
+
+    if torch.__version__ >= "2" and sys.platform != "win32":
+        model = torch.compile(model)
+
+    trainer.train(resume_from_checkpoint=None)
+
+    model.save_pretrained(output_dir)
+    trainer.save_model(output_dir)
+    
+    print(
+        "\n If there's a warning about missing keys above, please disregard :)"
+    )
+
+
+if __name__ == "__main__":
+    fire.Fire(train)