ros/LlamaFactory
1
0
Fork 0
mirror of https://github.com/hiyouga/LlamaFactory.git synced 2026-02-01 08:13:38 +00:00
Code Issues Packages Projects Releases Wiki Activity

refactor pissa, improve llamaboard

Browse Source 
Former-commit-id: 619556e46c19718f702c97df5d570a2a4c5fb13a
This commit is contained in:
hiyouga
2024-06-28 01:04:24 +08:00
parent edc7498111
commit 46f0189e88
16 changed files with 219 additions and 216 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/llamafactory/train/rm/workflow.py
View File

@@ -40,10 +40,9 @@
from typing import TYPE_CHECKING, List, Optional
from ...data import PairwiseDataCollatorWithPadding, get_dataset, split_dataset
from ...extras.callbacks import FixValueHeadModelCallback
from ...extras.misc import fix_valuehead_checkpoint
from ...extras.ploting import plot_loss
from ...model import load_model, load_tokenizer
from ..callbacks import fix_valuehead_checkpoint
from ..trainer_utils import create_modelcard_and_push
from .metric import compute_accuracy
from .trainer import PairwiseTrainer
@@ -77,7 +76,7 @@ def run_rm(
args=training_args,
finetuning_args=finetuning_args,
data_collator=data_collator,
callbacks=callbacks + [FixValueHeadModelCallback()],
callbacks=callbacks,
compute_metrics=compute_accuracy,
**tokenizer_module,
**split_dataset(dataset, data_args, training_args),
@@ -89,6 +88,7 @@ def run_rm(
trainer.save_model()
if training_args.should_save:
fix_valuehead_checkpoint(model, training_args.output_dir, training_args.save_safetensors)
trainer.log_metrics("train", train_result.metrics)
trainer.save_metrics("train", train_result.metrics)
trainer.save_state()