parser.py

import os
import sys
import torch
import datasets
import transformers
from typing import Any, Dict, Optional, Tuple
from transformers import HfArgumentParser
from transformers.trainer_utils import get_last_checkpoint

from utils.logging import get_logger
from args.data_args import DataArguments
from args.model_args import ModelArguments
from args.finetuning_args import FinetuningArguments
from args.training_args import TrainingArguments
from args.generation_args import GenerationArguments, InferArguments

logger = get_logger(__name__)

_TRAIN_ARGS = [
    ModelArguments, DataArguments, TrainingArguments, FinetuningArguments, GenerationArguments
]
_TRAIN_CLS = Tuple[
    ModelArguments, DataArguments, TrainingArguments, FinetuningArguments, GenerationArguments
]
_INFER_ARGS = [
    ModelArguments, DataArguments, TrainingArguments, FinetuningArguments, GenerationArguments, InferArguments
]
_INFER_CLS = Tuple[
    ModelArguments, DataArguments, TrainingArguments, FinetuningArguments, GenerationArguments, InferArguments
]



def parse_args(parser: "HfArgumentParser", args: Optional[Dict[str, Any]] = None) -> Tuple[Any]:
    if args is not None:
        return parser.parse_dict(args)
    elif len(sys.argv) == 2 and sys.argv[1].endswith(".yaml"):
        return parser.parse_yaml_file(os.path.abspath(sys.argv[1]))
    elif len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
        return parser.parse_json_file(os.path.abspath(sys.argv[1]))
    else:
        return parser.parse_args_into_dataclasses() 


def _verify_model_args(model_args: "ModelArguments", finetuning_args: "FinetuningArguments") -> None:
    if (
        model_args.checkpoint_dir is not None
        and len(model_args.checkpoint_dir) != 1
        and finetuning_args.finetuning_type != "lora"
    ):
        raise ValueError("Multiple checkpoints are only available for LoRA tuning.")


def parse_train_args(args: Optional[Dict[str, Any]] = None) -> _TRAIN_CLS:
    parser = HfArgumentParser(_TRAIN_ARGS)
    return parse_args(parser, args)


def parse_infer_args(args: Optional[Dict[str, Any]] = None) -> _INFER_CLS:
    parser = HfArgumentParser(_INFER_ARGS)
    return parse_args(parser, args)


def get_train_args(args: Optional[Dict[str, Any]] = None) -> _TRAIN_CLS:
    model_args, data_args, training_args, finetuning_args, generating_args = parse_train_args(args)

    # Setup logging
    if training_args.should_log:
        # The default of training_args.log_level is passive, so we set log level at info here to have that default.
        transformers.utils.logging.set_verbosity_info()

    log_level = training_args.get_process_log_level()
    datasets.utils.logging.set_verbosity(log_level)
    transformers.utils.logging.set_verbosity(log_level)
    transformers.utils.logging.enable_default_handler()
    transformers.utils.logging.enable_explicit_format()


    if finetuning_args.stage != "sft" and training_args.predict_with_generate:
        raise ValueError("`predict_with_generate` cannot be set as True except SFT.")

    if finetuning_args.stage == "sft" and training_args.do_predict and not training_args.predict_with_generate:
        raise ValueError("Please enable `predict_with_generate` to save model predictions.")

    if training_args.do_train and training_args.predict_with_generate:
        raise ValueError("`predict_with_generate` cannot be set as True while training.")
    

    _verify_model_args(model_args, finetuning_args)


    if training_args.do_train and (not training_args.fp16) and (not training_args.bf16):
        logger.warning("We recommend enable mixed precision training.")


    # postprocess training_args
    if (
        training_args.local_rank != -1
        and training_args.ddp_find_unused_parameters is None
        and finetuning_args.finetuning_type == "lora"
    ):
        logger.warning("`ddp_find_unused_parameters` needs to be set as False for LoRA in DDP training.")
        training_args_dict = training_args.to_dict()
        training_args_dict.update(dict(ddp_find_unused_parameters=False))
        training_args = TrainingArguments(**training_args_dict)

    if (
        training_args.resume_from_checkpoint is None
        and training_args.do_train
        and os.path.isdir(training_args.output_dir)
        and not training_args.overwrite_output_dir
    ):
        last_checkpoint = get_last_checkpoint(training_args.output_dir)
        if last_checkpoint is None and len(os.listdir(training_args.output_dir)) > 0:
            raise ValueError("Output directory already exists and is not empty. Please set `overwrite_output_dir`.")

        if last_checkpoint is not None:
            training_args_dict = training_args.to_dict()
            training_args_dict.update(dict(resume_from_checkpoint=last_checkpoint))
            training_args = TrainingArguments(**training_args_dict)
            logger.info("Resuming training from {}. Change `output_dir` or use `overwrite_output_dir` to avoid.".format(
                training_args.resume_from_checkpoint
            ))


    # postprocess model_args
    model_args.compute_dtype = (
        torch.bfloat16 if training_args.bf16 else (torch.float16 if training_args.fp16 else None)
    )
    model_args.model_max_length = data_args.cutoff_len

    # Log on each process the small summary:
    logger.info("Process rank: {}, device: {}, n_gpu: {}\n  distributed training: {}, compute dtype: {}".format(
        training_args.local_rank, training_args.device, training_args.n_gpu,
        bool(training_args.local_rank != -1), str(model_args.compute_dtype)
    ))
    logger.info(f"Training/evaluation parameters {training_args}")

    # Set seed before initializing model.
    transformers.set_seed(training_args.seed)

    return model_args, data_args, training_args, finetuning_args, generating_args



def get_infer_args(args: Optional[Dict[str, Any]] = None) -> _INFER_CLS:
    model_args, data_args, training_args, finetuning_args, generating_args, inference_args = parse_infer_args(args)

    if data_args.template is None:
        raise ValueError("Please specify which `template` to use.")

    _verify_model_args(model_args, finetuning_args)

    return model_args, data_args, training_args, finetuning_args, generating_args, inference_args