423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-23 22:32:54 +08:00
Code Issues Packages Projects Releases Wiki Activity

update

Browse Source 
Former-commit-id: f566ecc8d1f04615351acbe4f8480b75b2daed42
This commit is contained in:
Ting 2024-11-19 19:12:10 +08:00
parent bf2b8df540
commit 32656bc50d
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
src/llamafactory/train/dpo/workflow.py
View File

@ -66,10 +66,10 @@ def run_dpo(
# Update arguments # Update arguments
training_args.remove_unused_columns = False # important for multimodal and pairwise dataset training_args.remove_unused_columns = False # important for multimodal and pairwise dataset
effi_token_num = 0.0 effective_token_num = 0.0
for data in dataset_module["train_dataset"]: for data in dataset_module["train_dataset"]:
effi_token_num += len(data["chosen_input_ids"]) effective_token_num += len(data["chosen_input_ids"])
effi_token_num += len(data["rejected_input_ids"]) effective_token_num += len(data["rejected_input_ids"])
# Initialize our Trainer # Initialize our Trainer
trainer = CustomDPOTrainer( trainer = CustomDPOTrainer(
@ -87,7 +87,7 @@ def run_dpo(
if training_args.do_train: if training_args.do_train:
train_result = trainer.train(resume_from_checkpoint=training_args.resume_from_checkpoint) train_result = trainer.train(resume_from_checkpoint=training_args.resume_from_checkpoint)
train_result.metrics["effective_tokens_per_sec"] = ( train_result.metrics["effective_tokens_per_sec"] = (
effi_token_num * train_result.metrics["epoch"] / train_result.metrics["train_runtime"] effective_token_num * train_result.metrics["epoch"] / train_result.metrics["train_runtime"]
) )
if dist.is_initialized(): if dist.is_initialized():
train_result.metrics["effective_tokens_per_sec"] = ( train_result.metrics["effective_tokens_per_sec"] = (