Skip to content

fix: loss computation after embeddings resize - mllama #36840

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
zucchini-nlp merged 7 commits into from
Mar 21, 2025
Merged

fix: loss computation after embeddings resize - mllama #36840

Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
985fe75
move loss to generation class
Ssukriti Mar 19, 2025
98b8b29
code cleanup
Ssukriti Mar 19, 2025
dd02bfd
test for resize and loss computation
Ssukriti Mar 19, 2025
aef4f51
fix tests
Ssukriti Mar 20, 2025
c7b9adb
fix:test for resize and loss
Ssukriti Mar 20, 2025
75714c6
fix resize embedding mllama test
Ssukriti Mar 20, 2025
1d86452
review changes
Ssukriti Mar 20, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
fix tests 
Signed-off-by: Sukriti-Sharma4 <sukriti.sharma4@ibm.com>
  • Loading branch information
@Ssukriti
Ssukriti committed Mar 20, 2025
commit aef4f51596a1511e0ab0803086189e2b1a04b1e1
4 changes: 4 additions & 0 deletions tests/models/mllama/test_modeling_mllama.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -324,6 +324,9 @@ def test_inputs_embeds_matches_input_ids(self):
def test_resize_embeddings_results_in_successful_loss(self):
Copy link
Contributor Author

@Ssukriti Ssukriti Mar 20, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

test for reported bug that would fail earlier and now pass

# resizing embeddings should result in successful loss computation
config, inputs = self.model_tester.prepare_config_and_inputs_for_common()
labels = torch.zeros(
(self.batch_size, self.seq_length), dtype=torch.long, device=torch_device
)

for model_class in self.all_model_classes:
model = model_class(config)
Expand All @@ -334,6 +337,7 @@ def test_resize_embeddings_results_in_successful_loss(self):
loss = model(
input_ids=inputs["input_ids"],
attention_mask=inputs["attention_mask"],
labels=labels,
return_dict=True,
)["loss"]
self.parent.assertFalse(torch.isnan(loss).any().item())
Expand Down