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

Update no trainer scripts for language modeling and image classification examples (#18443)

Browse Source 
* Update no_trainer script for image-classification

* Update no_trainer scripts for language-modeling examples

* Remove unused variable

* Removing truncation from losses array for language modeling examples
This commit is contained in:
Ritik Nandwal
2022-08-03 18:03:18 +05:30
committed by GitHub
parent 10e1ec9a8c
commit 3db4378bd7
3 changed files with 3 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
examples/pytorch/image-classification/run_image_classification_no_trainer.py
View File

@@ -508,19 +508,11 @@ def main():
break break
model.eval() model.eval()
samples_seen = 0
for step, batch in enumerate(eval_dataloader): for step, batch in enumerate(eval_dataloader):
with torch.no_grad(): with torch.no_grad():
outputs = model(**batch) outputs = model(**batch)
predictions = outputs.logits.argmax(dim=-1) predictions = outputs.logits.argmax(dim=-1)
predictions, references = accelerator.gather((predictions, batch["labels"])) predictions, references = accelerator.gather_for_metrics((predictions, batch["labels"]))
# If we are in a multiprocess environment, the last batch has duplicates
if accelerator.num_processes > 1:
if step == len(eval_dataloader) - 1:
predictions = predictions[: len(eval_dataloader.dataset) - samples_seen]
references = references[: len(eval_dataloader.dataset) - samples_seen]
else:
samples_seen += references.shape[0]
metric.add_batch( metric.add_batch(
predictions=predictions, predictions=predictions,
references=references, references=references,

3
examples/pytorch/language-modeling/run_clm_no_trainer.py
View File

@@ -597,10 +597,9 @@ def main():
outputs = model(**batch) outputs = model(**batch)
loss = outputs.loss loss = outputs.loss
losses.append(accelerator.gather(loss.repeat(args.per_device_eval_batch_size))) losses.append(accelerator.gather_for_metrics(loss.repeat(args.per_device_eval_batch_size)))
losses = torch.cat(losses) losses = torch.cat(losses)
losses = losses[: len(eval_dataset)]
try: try:
eval_loss = torch.mean(losses) eval_loss = torch.mean(losses)
perplexity = math.exp(eval_loss) perplexity = math.exp(eval_loss)

3
examples/pytorch/language-modeling/run_mlm_no_trainer.py
View File

@@ -642,10 +642,9 @@ def main():
outputs = model(**batch) outputs = model(**batch)
loss = outputs.loss loss = outputs.loss
losses.append(accelerator.gather(loss.repeat(args.per_device_eval_batch_size))) losses.append(accelerator.gather_for_metrics(loss.repeat(args.per_device_eval_batch_size)))
losses = torch.cat(losses) losses = torch.cat(losses)
losses = losses[: len(eval_dataset)]
try: try:
eval_loss = torch.mean(losses) eval_loss = torch.mean(losses)
perplexity = math.exp(eval_loss) perplexity = math.exp(eval_loss)