Skip to content
This repository was archived by the owner on Oct 13, 2021. It is now read-only.

Fix GPT2 output order mismatch #476

Merged
merged 46 commits into from
May 7, 2020
Merged

Fix GPT2 output order mismatch #476

merged 46 commits into from
May 7, 2020

Conversation

jiafatom
Copy link
Collaborator

@jiafatom jiafatom commented May 7, 2020

No description provided.

jiafatom and others added 30 commits February 26, 2020 11:38
…nnx#394)

* Efficient-net test cases.

* Add DepthwiseConv2d to subclassed model.

* disable efficient net test cases.
@jiafatom jiafatom changed the title Fix GPT2 output order mismatch and add to transformer nightly build Fix GPT2 output order mismatch May 7, 2020
@wenbingl wenbingl merged commit ebc866d into onnx:master May 7, 2020
for model_instance_ in model_list:
keras.backend.clear_session()
model = model_instance_(config)
predictions = model.predict(inputs)
Copy link
Collaborator Author

Choose a reason for hiding this comment

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

This prediction works for transformers 2.5.0, but not for 2.8.0 with the errors:
> batch_element.shape, self.results.shape)) E ValueError: Mismatch between expected batch size and model output batch size. Output shape = (2, 1, 12, 10, 64), expected output shape = shape (1, 1, 12, 10, 64)

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants