SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Update trainer.md (#38113)

Browse Source 
Fix typo in torch.compile method parameters
This commit is contained in:
guspuffygit
2025-05-14 08:40:00 -04:00
committed by GitHub
parent 935bbbc711
commit 4a2decd192
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
docs/source/en/trainer.md
View File

@@ -372,14 +372,14 @@ accelerate launch \
### torch.compile ### torch.compile
[torch.compile](./perf_torch_compile) can significantly speed up training and reduce computational overhead. Configure your torch.compile settings in [`TrainingArguments`]. Set `torch.compile` to `True`, and select a backend and compile mode. [torch.compile](./perf_torch_compile) can significantly speed up training and reduce computational overhead. Configure your torch.compile settings in [`TrainingArguments`]. Set `torch_compile` to `True`, and select a backend and compile mode.
```py ```py
from transformers import TrainingArguments from transformers import TrainingArguments
training_args = TrainingArguments( training_args = TrainingArguments(
torch.compile=True, torch_compile=True,
torch.compile_backend="inductor", torch_compile_backend="inductor",
torch_compile_mode="default", torch_compile_mode="default",
..., ...,
) )