mirror of
https://github.com/hiyouga/LLaMA-Factory.git
synced 2025-11-05 18:32:14 +08:00
115 lines
4.2 KiB
Python
115 lines
4.2 KiB
Python
import os
|
|
import json
|
|
import torch
|
|
import numpy as np
|
|
import torch.nn as nn
|
|
from dataclasses import dataclass
|
|
from typing import Any, Dict, List, Optional, Sequence, Tuple, Union
|
|
|
|
from transformers.trainer import PredictionOutput
|
|
from transformers.tokenization_utils import PreTrainedTokenizer
|
|
|
|
import jieba
|
|
from rouge_chinese import Rouge
|
|
from nltk.translate.bleu_score import sentence_bleu, SmoothingFunction
|
|
|
|
from .peft_trainer import PeftTrainer
|
|
|
|
from .other import get_logger, IGNORE_INDEX
|
|
|
|
|
|
logger = get_logger(__name__)
|
|
|
|
|
|
@dataclass
|
|
class ComputeMetrics:
|
|
r"""
|
|
Wraps the tokenizer into metric functions, used in Seq2SeqPeftTrainer.
|
|
"""
|
|
|
|
tokenizer: PreTrainedTokenizer
|
|
|
|
def __call__(self, eval_preds: Sequence[Union[np.ndarray, Tuple[np.ndarray]]]) -> Dict[str, float]:
|
|
r"""
|
|
Uses the model predictions to compute metrics.
|
|
"""
|
|
preds, labels = eval_preds
|
|
score_dict = {"rouge-1": [], "rouge-2": [], "rouge-l": [], "bleu-4": []}
|
|
|
|
preds = np.where(preds != IGNORE_INDEX, preds, self.tokenizer.pad_token_id)
|
|
labels = np.where(labels != IGNORE_INDEX, labels, self.tokenizer.pad_token_id)
|
|
|
|
decoded_preds = self.tokenizer.batch_decode(preds, skip_special_tokens=True)
|
|
decoded_labels = self.tokenizer.batch_decode(labels, skip_special_tokens=True)
|
|
|
|
for pred, label in zip(decoded_preds, decoded_labels):
|
|
hypothesis = list(jieba.cut(pred))
|
|
reference = list(jieba.cut(label))
|
|
|
|
if len(" ".join(hypothesis).split()) == 0:
|
|
result = {"rouge-1": {"f": 0.0}, "rouge-2": {"f": 0.0}, "rouge-l": {"f": 0.0}}
|
|
else:
|
|
rouge = Rouge()
|
|
scores = rouge.get_scores(" ".join(hypothesis), " ".join(reference))
|
|
result = scores[0]
|
|
|
|
for k, v in result.items():
|
|
score_dict[k].append(round(v["f"] * 100, 4))
|
|
|
|
bleu_score = sentence_bleu([list(label)], list(pred), smoothing_function=SmoothingFunction().method3)
|
|
score_dict["bleu-4"].append(round(bleu_score * 100, 4))
|
|
|
|
return {k: float(np.mean(v)) for k, v in score_dict.items()}
|
|
|
|
|
|
class Seq2SeqPeftTrainer(PeftTrainer):
|
|
r"""
|
|
Inherits PeftTrainer to compute generative metrics such as BLEU and ROUGE.
|
|
"""
|
|
|
|
def prediction_step(
|
|
self,
|
|
model: nn.Module,
|
|
inputs: Dict[str, Union[torch.Tensor, Any]],
|
|
prediction_loss_only: bool,
|
|
ignore_keys: Optional[List[str]] = None,
|
|
) -> Tuple[Optional[float], Optional[torch.Tensor], Optional[torch.Tensor]]:
|
|
r"""
|
|
Removes the prompt part in the generated tokens.
|
|
|
|
Subclass and override to inject custom behavior.
|
|
"""
|
|
input_ids = inputs["input_ids"]
|
|
loss, generated_tokens, labels = super().prediction_step(
|
|
model, inputs, prediction_loss_only=prediction_loss_only, ignore_keys=ignore_keys
|
|
)
|
|
generated_tokens = generated_tokens[:, input_ids.size(-1):] if generated_tokens is not None else None
|
|
return (loss, generated_tokens, labels)
|
|
|
|
def save_predictions(
|
|
self,
|
|
predict_results: PredictionOutput
|
|
) -> None:
|
|
r"""
|
|
Saves model predictions to `output_dir`.
|
|
|
|
A custom behavior that not contained in Seq2SeqTrainer.
|
|
"""
|
|
if not self.is_world_process_zero():
|
|
return
|
|
|
|
output_prediction_file = os.path.join(self.args.output_dir, "generated_predictions.jsonl")
|
|
logger.info(f"Saving prediction results to {output_prediction_file}")
|
|
|
|
preds = np.where(predict_results.predictions != IGNORE_INDEX, predict_results.predictions, self.tokenizer.pad_token_id)
|
|
labels = np.where(predict_results.label_ids != IGNORE_INDEX, predict_results.label_ids, self.tokenizer.pad_token_id)
|
|
|
|
decoded_preds = self.tokenizer.batch_decode(preds, skip_special_tokens=True)
|
|
decoded_labels = self.tokenizer.batch_decode(labels, skip_special_tokens=True)
|
|
|
|
with open(output_prediction_file, "w", encoding="utf-8") as writer:
|
|
res: List[str] = []
|
|
for pred, label in zip(decoded_preds, decoded_labels):
|
|
res.append(json.dumps({"label": label, "predict": pred}, ensure_ascii=False))
|
|
writer.write("\n".join(res))
|