Repository navigation
[ci] fix compilation CI. - #15092
[ci] fix compilation CI.#15092sayakpaul wants to merge 2 commits into
Conversation
| "time_dim": 16, | ||
| "patch_size": (1, 2, 2), | ||
| "model_dim": 24, | ||
| "model_dim": 32, |
There was a problem hiding this comment.
To satisfy the requirements of Flex + compile.
| raises=(torch._dynamo.exc.Unsupported, torch._dynamo.exc.RecompileError), | ||
| strict=True, | ||
| ) | ||
| def test_compile_on_different_shapes(self): |
There was a problem hiding this comment.
We can make it work with some modeling-level changes but not a priority for this model at the moment IMO.
|
/diffusers-bot pytest tests/models/autoencoders/test_models_autoencoder_mmaudio.py tests/models/autoencoders/test_models_kandinsky6_vocoder.py tests/models/latent_upscaler/test_models_kandinsky6_latent_upscaler.py tests/models/transformers/test_models_transformer_kandinsky6_sr.py |
|
✅ |
| pass | ||
| @pytest.mark.xfail( | ||
| condition=is_torch_version("<", "2.12"), | ||
| reason="torch < 2.12 cannot trace `AttentionBackendName(backend)` in `dispatch_attention_fn` in a full graph.", |
There was a problem hiding this comment.
Can't we fix this in dispatch_attention_fn? Our min supported version is torch 2.6 so if this can be addressed in src, we should handle it.
| @is_torch_compile | ||
| @require_accelerator | ||
| @require_accelerate | ||
| class BFLCompileTesterMixin(BFLConfigMixin, QuantizationCompileTesterMixin): |
There was a problem hiding this comment.
Instead of a helper method, why not add a decoator here:
@pytest.mark.skipif(
torch_device == "cuda" and torch.cuda.get_device_capability() < (8, 9),
reason="torch.compile of BFL models needs FP8 support in Triton (compute capability 8.9+)",
)
class BFLCompileTesterMixin(BFLConfigMixin, QuantizationCompileTesterMixin):This branch has not been deployed
2 participants
Fixes https://github.com/huggingface/diffusers/actions/runs/39450097188/job/118722310713