ros/LLaMA-Factory
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

format style

Browse Source 
Former-commit-id: 53b683531b83cd1d19de97c6565f16c1eca6f5e1
This commit is contained in:
hiyouga
2024-01-20 20:15:56 +08:00
parent 1750218057
commit 66e0e651b9
73 changed files with 1492 additions and 2325 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
src/llmtuner/train/sft/metric.py
View File

@@ -1,11 +1,11 @@
import numpy as np
from dataclasses import dataclass
from typing import TYPE_CHECKING, Dict, Sequence, Tuple, Union
import numpy as np
from ...extras.constants import IGNORE_INDEX
from ...extras.packages import (
is_jieba_available, is_nltk_available, is_rouge_available
)
from ...extras.packages import is_jieba_available, is_nltk_available, is_rouge_available
if TYPE_CHECKING:
from transformers.tokenization_utils import PreTrainedTokenizer
@@ -14,7 +14,7 @@ if is_jieba_available():
import jieba
if is_nltk_available():
from nltk.translate.bleu_score import sentence_bleu, SmoothingFunction
from nltk.translate.bleu_score import SmoothingFunction, sentence_bleu
if is_rouge_available():
from rouge_chinese import Rouge

47
src/llmtuner/train/sft/trainer.py
View File

@@ -1,14 +1,16 @@
import os
import json
import torch
import numpy as np
import torch.nn as nn
import os
from typing import TYPE_CHECKING, Any, Dict, List, Optional, Tuple, Union
import numpy as np
import torch
import torch.nn as nn
from transformers import Seq2SeqTrainer
from ...extras.constants import IGNORE_INDEX
from ...extras.logging import get_logger
if TYPE_CHECKING:
from transformers.trainer import PredictionOutput
@@ -33,16 +35,16 @@ class CustomSeq2SeqTrainer(Seq2SeqTrainer):
Subclass and override to inject custom behavior.
"""
labels = inputs["labels"].detach().clone() if "labels" in inputs else None # backup labels
labels = inputs["labels"].detach().clone() if "labels" in inputs else None # backup labels
if self.args.predict_with_generate:
assert self.tokenizer.padding_side == "left", "This method only accepts left-padded tensor."
prompt_len, label_len = inputs["input_ids"].size(-1), inputs["labels"].size(-1)
if prompt_len > label_len:
inputs["labels"] = self._pad_tensors_to_target_len(inputs["labels"], inputs["input_ids"])
if label_len > prompt_len: # truncate the labels instead of padding the inputs (llama2 fp16 compatibility)
if label_len > prompt_len: # truncate the labels instead of padding the inputs (llama2 fp16 compatibility)
inputs["labels"] = inputs["labels"][:, :prompt_len]
loss, generated_tokens, _ = super().prediction_step( # ignore the returned labels (may be truncated)
loss, generated_tokens, _ = super().prediction_step( # ignore the returned labels (may be truncated)
model, inputs, prediction_loss_only=prediction_loss_only, ignore_keys=ignore_keys
)
if generated_tokens is not None and self.args.predict_with_generate:
@@ -51,23 +53,16 @@ class CustomSeq2SeqTrainer(Seq2SeqTrainer):
return loss, generated_tokens, labels
def _pad_tensors_to_target_len(
self,
src_tensor: torch.Tensor,
tgt_tensor: torch.Tensor
) -> torch.Tensor:
def _pad_tensors_to_target_len(self, src_tensor: torch.Tensor, tgt_tensor: torch.Tensor) -> torch.Tensor:
r"""
Pads the tensor to the same length as the target tensor.
"""
assert self.tokenizer.pad_token_id is not None, "Pad token is required."
padded_tensor = self.tokenizer.pad_token_id * torch.ones_like(tgt_tensor)
padded_tensor[:, -src_tensor.shape[-1]:] = src_tensor # adopt left-padding
return padded_tensor.contiguous() # in contiguous memory
padded_tensor[:, -src_tensor.shape[-1] :] = src_tensor # adopt left-padding
return padded_tensor.contiguous() # in contiguous memory
def save_predictions(
self,
predict_results: "PredictionOutput"
) -> None:
def save_predictions(self, predict_results: "PredictionOutput") -> None:
r"""
Saves model predictions to `output_dir`.
@@ -79,15 +74,23 @@ class CustomSeq2SeqTrainer(Seq2SeqTrainer):
output_prediction_file = os.path.join(self.args.output_dir, "generated_predictions.jsonl")
logger.info(f"Saving prediction results to {output_prediction_file}")
labels = np.where(predict_results.label_ids != IGNORE_INDEX, predict_results.label_ids, self.tokenizer.pad_token_id)
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
)
preds = np.where(
predict_results.predictions != IGNORE_INDEX, predict_results.predictions, self.tokenizer.pad_token_id
)
for i in range(len(preds)):
pad_len = np.nonzero(preds[i] != self.tokenizer.pad_token_id)[0]
if len(pad_len):
preds[i] = np.concatenate((preds[i][pad_len[0]:], preds[i][:pad_len[0]]), axis=-1) # move pad token to last
preds[i] = np.concatenate(
(preds[i][pad_len[0] :], preds[i][: pad_len[0]]), axis=-1
) # move pad token to last
decoded_labels = self.tokenizer.batch_decode(labels, skip_special_tokens=True, clean_up_tokenization_spaces=False)
decoded_labels = self.tokenizer.batch_decode(
labels, skip_special_tokens=True, clean_up_tokenization_spaces=False
)
decoded_preds = self.tokenizer.batch_decode(preds, skip_special_tokens=True, clean_up_tokenization_spaces=True)
with open(output_prediction_file, "w", encoding="utf-8") as writer:

32
src/llmtuner/train/sft/workflow.py
View File

@@ -1,6 +1,7 @@
# Inspired by: https://github.com/huggingface/transformers/blob/v4.34.1/examples/pytorch/summarization/run_summarization.py
from typing import TYPE_CHECKING, Optional, List
from typing import TYPE_CHECKING, List, Optional
from transformers import DataCollatorForSeq2Seq, Seq2SeqTrainingArguments
from ...data import get_dataset, split_dataset
@@ -15,7 +16,8 @@ from ...train.utils import create_modelcard_and_push
if TYPE_CHECKING:
from transformers import TrainerCallback
from ...hparams import ModelArguments, DataArguments, FinetuningArguments, GeneratingArguments
from ...hparams import DataArguments, FinetuningArguments, GeneratingArguments, ModelArguments
def run_sft(
@@ -24,29 +26,31 @@ def run_sft(
training_args: "Seq2SeqTrainingArguments",
finetuning_args: "FinetuningArguments",
generating_args: "GeneratingArguments",
callbacks: Optional[List["TrainerCallback"]] = None
callbacks: Optional[List["TrainerCallback"]] = None,
):
model, tokenizer = load_model_and_tokenizer(model_args, finetuning_args, training_args.do_train)
dataset = get_dataset(tokenizer, model_args, data_args, training_args, stage="sft")
if training_args.predict_with_generate:
tokenizer.padding_side = "left" # use left-padding in generation
tokenizer.padding_side = "left" # use left-padding in generation
if getattr(model, "is_quantized", False) and not training_args.do_train:
setattr(model, "_hf_peft_config_loaded", True) # hack here: make model compatible with prediction
setattr(model, "_hf_peft_config_loaded", True) # hack here: make model compatible with prediction
data_collator = DataCollatorForSeq2Seq(
tokenizer=tokenizer,
pad_to_multiple_of=8 if tokenizer.padding_side == "right" else None, # for shift short attention
label_pad_token_id=IGNORE_INDEX if data_args.ignore_pad_token_for_loss else tokenizer.pad_token_id
pad_to_multiple_of=8 if tokenizer.padding_side == "right" else None, # for shift short attention
label_pad_token_id=IGNORE_INDEX if data_args.ignore_pad_token_for_loss else tokenizer.pad_token_id,
)
# Override the decoding parameters of Seq2SeqTrainer
training_args_dict = training_args.to_dict()
training_args_dict.update(dict(
generation_max_length=training_args.generation_max_length or data_args.cutoff_len,
generation_num_beams=data_args.eval_num_beams or training_args.generation_num_beams
))
training_args_dict.update(
dict(
generation_max_length=training_args.generation_max_length or data_args.cutoff_len,
generation_num_beams=data_args.eval_num_beams or training_args.generation_num_beams,
)
)
training_args = Seq2SeqTrainingArguments(**training_args_dict)
# Initialize our Trainer
@@ -57,7 +61,7 @@ def run_sft(
data_collator=data_collator,
callbacks=callbacks,
compute_metrics=ComputeMetrics(tokenizer) if training_args.predict_with_generate else None,
**split_dataset(dataset, data_args, training_args)
**split_dataset(dataset, data_args, training_args),
)
# Keyword arguments for `model.generate`
@@ -79,7 +83,7 @@ def run_sft(
# Evaluation
if training_args.do_eval:
metrics = trainer.evaluate(metric_key_prefix="eval", **gen_kwargs)
if training_args.predict_with_generate: # eval_loss will be wrong if predict_with_generate is enabled
if training_args.predict_with_generate: # eval_loss will be wrong if predict_with_generate is enabled
metrics.pop("eval_loss", None)
trainer.log_metrics("eval", metrics)
trainer.save_metrics("eval", metrics)
@@ -87,7 +91,7 @@ def run_sft(
# Predict
if training_args.do_predict:
predict_results = trainer.predict(dataset, metric_key_prefix="predict", **gen_kwargs)
if training_args.predict_with_generate: # predict_loss will be wrong if predict_with_generate is enabled
if training_args.predict_with_generate: # predict_loss will be wrong if predict_with_generate is enabled
predict_results.metrics.pop("predict_loss", None)
trainer.log_metrics("predict", predict_results.metrics)
trainer.save_metrics("predict", predict_results.metrics)