|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
from typing import TYPE_CHECKING, List, Optional |
|
|
|
from ...data import SFTDataCollatorWith4DAttentionMask, get_dataset, split_dataset |
|
from ...extras.constants import IGNORE_INDEX |
|
from ...extras.misc import get_logits_processor |
|
from ...extras.ploting import plot_loss |
|
from ...model import load_model, load_tokenizer |
|
from ..trainer_utils import create_modelcard_and_push |
|
from .metric import ComputeMetrics, compute_accuracy, eval_logit_processor |
|
from .trainer import CustomSeq2SeqTrainer |
|
|
|
|
|
if TYPE_CHECKING: |
|
from transformers import Seq2SeqTrainingArguments, TrainerCallback |
|
|
|
from ...hparams import DataArguments, FinetuningArguments, GeneratingArguments, ModelArguments |
|
|
|
|
|
def run_sft( |
|
model_args: "ModelArguments", |
|
data_args: "DataArguments", |
|
training_args: "Seq2SeqTrainingArguments", |
|
finetuning_args: "FinetuningArguments", |
|
generating_args: "GeneratingArguments", |
|
callbacks: Optional[List["TrainerCallback"]] = None, |
|
): |
|
tokenizer_module = load_tokenizer(model_args) |
|
tokenizer = tokenizer_module["tokenizer"] |
|
dataset = get_dataset(model_args, data_args, training_args, stage="sft", **tokenizer_module) |
|
model = load_model(tokenizer, model_args, finetuning_args, training_args.do_train) |
|
|
|
if training_args.predict_with_generate: |
|
tokenizer.padding_side = "left" |
|
|
|
if getattr(model, "is_quantized", False) and not training_args.do_train: |
|
setattr(model, "_hf_peft_config_loaded", True) |
|
|
|
data_collator = SFTDataCollatorWith4DAttentionMask( |
|
tokenizer=tokenizer, |
|
pad_to_multiple_of=8 if tokenizer.padding_side == "right" else None, |
|
label_pad_token_id=IGNORE_INDEX if data_args.ignore_pad_token_for_loss else tokenizer.pad_token_id, |
|
block_diag_attn=model_args.block_diag_attn, |
|
attn_implementation=getattr(model.config, "_attn_implementation", None), |
|
compute_dtype=model_args.compute_dtype, |
|
) |
|
|
|
|
|
training_args.generation_max_length = training_args.generation_max_length or data_args.cutoff_len |
|
training_args.generation_num_beams = data_args.eval_num_beams or training_args.generation_num_beams |
|
training_args.remove_unused_columns = False if model_args.visual_inputs else training_args.remove_unused_columns |
|
|
|
|
|
trainer = CustomSeq2SeqTrainer( |
|
model=model, |
|
args=training_args, |
|
finetuning_args=finetuning_args, |
|
data_collator=data_collator, |
|
callbacks=callbacks, |
|
compute_metrics=ComputeMetrics(tokenizer) if training_args.predict_with_generate else compute_accuracy, |
|
preprocess_logits_for_metrics=None if training_args.predict_with_generate else eval_logit_processor, |
|
**tokenizer_module, |
|
**split_dataset(dataset, data_args, training_args), |
|
) |
|
|
|
|
|
gen_kwargs = generating_args.to_dict() |
|
gen_kwargs["eos_token_id"] = [tokenizer.eos_token_id] + tokenizer.additional_special_tokens_ids |
|
gen_kwargs["pad_token_id"] = tokenizer.pad_token_id |
|
gen_kwargs["logits_processor"] = get_logits_processor() |
|
|
|
|
|
if training_args.do_train: |
|
train_result = trainer.train(resume_from_checkpoint=training_args.resume_from_checkpoint) |
|
trainer.save_model() |
|
trainer.log_metrics("train", train_result.metrics) |
|
trainer.save_metrics("train", train_result.metrics) |
|
trainer.save_state() |
|
if trainer.is_world_process_zero() and finetuning_args.plot_loss: |
|
plot_loss(training_args.output_dir, keys=["loss", "eval_loss", "eval_accuracy"]) |
|
|
|
|
|
if training_args.do_eval: |
|
metrics = trainer.evaluate(metric_key_prefix="eval", **gen_kwargs) |
|
if training_args.predict_with_generate: |
|
metrics.pop("eval_loss", None) |
|
trainer.log_metrics("eval", metrics) |
|
trainer.save_metrics("eval", metrics) |
|
|
|
|
|
if training_args.do_predict: |
|
predict_results = trainer.predict(dataset, metric_key_prefix="predict", **gen_kwargs) |
|
if training_args.predict_with_generate: |
|
predict_results.metrics.pop("predict_loss", None) |
|
trainer.log_metrics("predict", predict_results.metrics) |
|
trainer.save_metrics("predict", predict_results.metrics) |
|
trainer.save_predictions(dataset, predict_results) |
|
|
|
|
|
create_modelcard_and_push(trainer, model_args, data_args, training_args, finetuning_args) |
|
|