Skip to content

Fix TE 2.12 AllGather CI failure #3101

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
yanring merged 4 commits into from
Feb 3, 2026
Merged

Fix TE 2.12 AllGather CI failure #3101

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
626c78d
fix TE 2.12 AllGather CI failure
BestJuly Jan 27, 2026
1c5ff56
Merge branch 'main' into lit/fix_ci_0127
BestJuly Jan 27, 2026
7859d3b
re-enable previous UT
BestJuly Jan 27, 2026
96f5242
Merge branch 'main' into lit/fix_ci_0127
jaredcasper Feb 3, 2026
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
2 changes: 1 addition & 1 deletion megatron/core/transformer/moe/token_dispatcher.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -297,7 +297,7 @@ def dispatch_postprocess(self, hidden_states, probs):
(permuted_local_hidden_states, _, self.reversed_local_input_permutation_mapping) = permute(
hidden_states,
self.local_map,
num_out_tokens=tokens_per_expert.sum(),
num_out_tokens=tokens_per_expert.sum().item(),
Comment thread
yaox12 marked this conversation as resolved.
fused=self.config.moe_permute_fusion,
)

Expand Down
1 change: 0 additions & 1 deletion tests/unit_tests/transformer/moe/test_a2a_token_dispatcher.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -95,7 +95,6 @@ def test_capacity_padding_forward_backward(self, tp_size, ep_size, permute_fusio
)
container.dispatcher_drop_and_pad_test()

@pytest.mark.flaky_in_dev
@pytest.mark.skipif(
not is_te_min_version("1.7.0"), reason="TE 1.7.0 is required for MoE with FP8."
)
Expand Down
2 changes: 0 additions & 2 deletions tests/unit_tests/transformer/moe/test_token_dispatcher.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -364,7 +364,6 @@ def setup_method(self, method):
def teardown_method(self, method):
Utils.destroy_model_parallel()

@pytest.mark.flaky_in_dev
@pytest.mark.skipif(not torch.cuda.is_available(), reason="CUDA not available")
@pytest.mark.internal
@pytest.mark.parametrize("tp_size,ep_size", [(8, 1), (1, 8), (2, 4), (1, 1)])
Expand All @@ -383,7 +382,6 @@ def test_forward_backward(self, tp_size, ep_size, permute_fusion):

container.dispatcher_dropless_test()

@pytest.mark.flaky_in_dev
@pytest.mark.skipif(not torch.cuda.is_available(), reason="CUDA not available")
@pytest.mark.internal
@pytest.mark.parametrize("permute_fusion", permute_fusion_params)
Expand Down
Loading