Skip to content

Fix WhisperModelTest #21883

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
ydshieh merged 2 commits into from
Mar 1, 2023
Merged

Fix WhisperModelTest #21883

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e571b04
force on the same device
ydshieh Mar 1, 2023
14a48c4
fix tests
ydshieh Mar 1, 2023
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
14 changes: 13 additions & 1 deletion tests/models/whisper/test_modeling_whisper.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -284,6 +284,8 @@ class WhisperModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMi
fx_compatible = False
test_pruning = False
test_missing_keys = False
# Needs higher percentages after model tester's vocab_size is changed to 200 (PR #21222)
model_split_percents = [0.8, 0.9]
Copy link
Collaborator Author

@ydshieh ydshieh Mar 1, 2023

Choose a reason for hiding this comment

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

To fix test_model_parallelism


input_name = "input_features"

Expand Down Expand Up @@ -727,7 +729,17 @@ def _create_and_check_torchscript(self, config, inputs_dict):
input_features = inputs["input_features"]
decoder_input_ids = inputs["decoder_input_ids"]
decoder_attention_mask = inputs["decoder_attention_mask"]
traced_model = torch.jit.trace(model, (input_features, decoder_input_ids, decoder_attention_mask))
# prepare `attention_mask` with shape (batch_size, sequence_length)
Copy link
Collaborator Author

@ydshieh ydshieh Mar 1, 2023

Choose a reason for hiding this comment

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

Due to the change in #21298 (a new optional argument being added)

attention_mask = torch.ones(
input_features.shape[0],
input_features.shape[-1],
device=input_features.device,
dtype=input_features.dtype,
)
Copy link
Collaborator Author

@ydshieh ydshieh Mar 1, 2023

Choose a reason for hiding this comment

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

I am a bit lazy, so I just prepare this argument here directly.

traced_model = torch.jit.trace(
model, (input_features, attention_mask, decoder_input_ids, decoder_attention_mask)
)

except RuntimeError:
self.fail("Couldn't trace module.")

Expand Down