Mamba: add generative tests #31478
Merged
Mamba: add generative tests #31478
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
gante
commented
Jun 18, 2024
| @@ -250,6 +250,8 @@ def prepare_config_and_inputs_for_common(self): | |||
| @require_torch | |||
| class MambaModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin, unittest.TestCase): | |||
| all_model_classes = (MambaModel, MambaForCausalLM) if is_torch_available() else () | |||
| all_generative_model_classes = (MambaForCausalLM,) if is_torch_available() else () | |||
There was a problem hiding this comment.
all tests passing for mamba :)
gante
commented
Jun 18, 2024
Comment on lines
-295
to
-297
| @unittest.skip("No attention in mamba") | ||
| def test_retain_grad_hidden_states_attentions(self): | ||
| pass |
There was a problem hiding this comment.
this test was actually passing, just needed the has_attentions flag
gante
commented
Jun 18, 2024
Comment on lines
-367
to
-373
| @unittest.skip("Mamba does not use attention") | ||
| def test_attention_outputs(self): | ||
| r""" | ||
| Overriding the test_attention_outputs test as the attention outputs of Mamba are different from other models | ||
| it has a shape `batch_size, seq_len, hidden_size`. | ||
| """ | ||
| pass |
There was a problem hiding this comment.
has_attentions skips this test
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
amyeroberts
approved these changes
Jun 18, 2024
zucchini-nlp
approved these changes
Jun 19, 2024
This was referenced Jun 19, 2024
itazap
pushed a commit
that referenced
this pull request
Jun 20, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Supercedes #31094
Fixes #30828 🤗
This PR:
mambahas_attentionstester flag (TIL about this flag)PreTrainedModel--_is_stateful(self explanatory) [note: AFAIK we only have 3 stateful models,mamba,jamba, andrwkv, lmk if there are more!]_is_statefulto set a few appropriate exceptions and test skipsNote: with
has_attentionsand_is_stateful, I suspect I can remove a few more skips in a follow up PR :)