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

Fix metric calculation in examples and setup tests to run on multi-gpu for no_trainer scripts (#17331)

Browse Source 
* Fix length in no_trainer examples

* Add setup and teardown

* Use new accelerator config generator to automatically make tests able to run based on environment
This commit is contained in:
Zachary Mueller
2022-05-18 14:17:40 -04:00
committed by GitHub
parent 6e195eb9de
commit 1762ded30a
8 changed files with 91 additions and 121 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
examples/pytorch/translation/run_translation_no_trainer.py
View File

@@ -661,11 +661,11 @@ def main():
# If we are in a multiprocess environment, the last batch has duplicates
if accelerator.num_processes > 1:
if step == len(eval_dataloader):
if step == len(eval_dataloader) - 1:
decoded_preds = decoded_preds[: len(eval_dataloader.dataset) - samples_seen]
decoded_labels = decoded_labels[: len(eval_dataloader.dataset) - samples_seen]
else:
samples_seen += decoded_labels.shape[0]
samples_seen += len(decoded_labels)
metric.add_batch(predictions=decoded_preds, references=decoded_labels)
eval_metric = metric.compute()