finetune.py

import os
import sys
from typing import List
import json

import torch
import transformers
from datasets import load_dataset

"""
Unused imports:
import torch.nn as nn
import bitsandbytes as bnb

"""

assert (
    "LlamaTokenizer" in transformers._import_structure["models.llama"]
), "LLaMA is now in HuggingFace's main branch.\nPlease reinstall it: pip uninstall transformers && pip install git+https://github.com/huggingface/transformers.git"

from peft import (
    LoraConfig,
    get_peft_model,
    get_peft_model_state_dict,
    prepare_model_for_int8_training,
    set_peft_model_state_dict,
)
from transformers import LlamaForCausalLM, LlamaTokenizer

from utils.prompter import Prompter

from transformers import AutoModelForCausalLM, AutoConfig, AutoModelForCausalLM
from accelerate import init_empty_weights, infer_auto_device_map
import torch
from torch.utils.tensorboard import SummaryWriter
writer = SummaryWriter()

# quantization_config = BitsAndBytesConfig(llm_int8_enable_fp32_cpu_offload=True)

import argparse


def train(
    # model/data params
    base_model: str = "",  # the only required argument
    data_path: str = "",
    output_dir: str = "",
    # training hyperparams
    batch_size: int = 128,
    micro_batch_size: int = 4,
    num_epochs: int = 1,
    learning_rate: float = 3e-4,
    cutoff_len: int = 512,
    val_set_size: int = 2000,
    # lora hyperparams
    lora_r: int = 8,
    lora_alpha: int = 16,
    lora_dropout: float = 0.05,
    lora_target_modules: List[str] = [
        "q_proj",
        "v_proj",
    ],
    # llm hyperparams
    train_on_inputs: bool = True,  # 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
    
    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.
    cuda_id: int = 0,
):
    if int(os.environ.get("LOCAL_RANK", 0)) == 0:
        print(
            f"Training Model by LoRA 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"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"add_eos_token: {add_eos_token}\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"
        )
    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
            
    do_int8 = False

    with init_empty_weights():
        config = AutoConfig.from_pretrained(base_model)
        model = AutoModelForCausalLM.from_config(config)

    # d = {1:"24GiB", 2:"24GiB"}
    d = {cuda_id: "24GiB"}
    for i in range(0, 4):
        d[i] = "24GiB"
    device_map = infer_auto_device_map(
        model, max_memory=d, dtype=torch.int8 if do_int8 else torch.float16,
        no_split_module_classes=["BloomBlock", "OPTDecoderLayer", "LlamaDecoderLayer"]
    )
    print(device_map)

    model = LlamaForCausalLM.from_pretrained(
        base_model,
        load_in_8bit=True,
        torch_dtype=torch.float16,
        device_map=device_map,
    )

    tokenizer = LlamaTokenizer.from_pretrained(
        base_model
    )

    tokenizer.pad_token_id = (
        0  # unk. we want this to be different from the eos token
    )
    tokenizer.padding_side = "left"  # Allow batched inference

    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)
    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")

    print("***** Running training *****")
    print(f"  Num Epochs = {num_epochs}", )
    print(f"  Instantaneous batch size per GPU = {micro_batch_size}")
    print(f"  Gradient Accumulation steps = {gradient_accumulation_steps}")
    print(f"  Total train batch size (w. parallel, distributed & accumulation) = {batch_size}")
    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=100,
            num_train_epochs=num_epochs,
            learning_rate=learning_rate,
            fp16=True,
            logging_steps=10,
            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,
            output_dir=output_dir,
            save_total_limit=3,
            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,
        ),
        data_collator=transformers.DataCollatorForSeq2Seq(
            tokenizer, pad_to_multiple_of=8, return_tensors="pt", padding=True
        ),
    )
    model.config.use_cache = False

    # Seemly this piece of code has shown the issues of save model as 443B
    # 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=resume_from_checkpoint)

    model.save_pretrained(output_dir)

    print(
        "\n If there's a warning about missing keys above, please disregard :)"
    )


if __name__ == "__main__":
    parser = argparse.ArgumentParser(description='Process some integers.')
    parser.add_argument('--base_model', type=str, default='/home/daven/llm/qokori/llama-2023-05-27-23-10/checkpoint-27400/', help='the based model used for Peft')
    parser.add_argument('--data_path', type=str, default='/home/daven/llm/qokori/qokori-sft/sft_data_clean/geosignal_sft/data/alpaca_data_gpt4.json', help='the data used for instructing tuning')
    parser.add_argument('--output_dir', type=str, default='/home/daven/llm/qokori/qokori-sft/outputs/geo_llama_alpacagpt4/', help='output model')
    parser.add_argument('--cuda_id', type=int, default=0, help='cuda nvidia-smi id')
    parser.add_argument('--resume_from_checkpoint', type=str, default=None, help='resume_from_checkpoint')
    parser.add_argument('--num_epochs', type=int, default=1, help='num_epochs')
    parser.add_argument('--lora_target_modules', nargs='+', default=["q_proj", "k_proj"], help='''model module, e.g. ["q_proj", "k_proj", "v_proj", "o_proj", "down_proj", "up_proj"]''')
    

    args = parser.parse_args()
    
    print(f"Lora Target modules:\n{args.lora_target_modules}")
    
    train(
        base_model= args.base_model,
        data_path = args.data_path,
        output_dir = args.output_dir,
        cuda_id=args.cuda_id,
        num_epochs=args.num_epochs,
        lora_target_modules=args.lora_target_modules,
        resume_from_checkpoint=args.resume_from_checkpoint
    )