Compare commits

..

1 Commits

Author SHA1 Message Date
NanoCode012
a65dbe779f fix: suspected eval vram increased usage 2025-06-23 18:44:03 +07:00
2 changed files with 3 additions and 4 deletions

View File

@@ -215,10 +215,6 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
self.cfg.auto_find_batch_size self.cfg.auto_find_batch_size
) )
training_arguments_kwargs["eval_accumulation_steps"] = (
self.cfg.gradient_accumulation_steps
)
training_arguments_kwargs["load_best_model_at_end"] = ( training_arguments_kwargs["load_best_model_at_end"] = (
( (
self.cfg.load_best_model_at_end is not False self.cfg.load_best_model_at_end is not False

View File

@@ -74,6 +74,9 @@ def is_main_process() -> bool:
Check if the current process is the main process. If not in distributed mode, Check if the current process is the main process. If not in distributed mode,
always return `True`. always return `True`.
We use a simpler logic when the distributed state is not initialized: we just log
on the 0-th local rank.
Returns: Returns:
`True` if the current process is the main process, `False` otherwise. `True` if the current process is the main process, `False` otherwise.
""" """