kanshan/TensorRT-LLMs
1
0
Fork 0
mirror of https://github.com/NVIDIA/TensorRT-LLM.git synced 2026-01-14 06:27:45 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

[https://nvbugs/5575841] [test] Move test_moe.py to serial tests to improve stability + unwaive FP4 MoE torch unit tests (#8422)

Browse Source 
Signed-off-by: Dom Brown <3886319+DomBrown@users.noreply.github.com>
This commit is contained in:
Dom Brown 2025-10-30 12:57:56 +00:00 committed by GitHub
parent 0dac57f2bc
commit 9410ce3bea
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 0 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
tests/unittest/_torch/thop/parallel/test_moe.py → tests/unittest/_torch/thop/serial/test_moe.py
View File

@ -991,7 +991,6 @@ class TestMoeFp4:
the default tactic selection works. This reduces unnecessary test runs for CI
"""
@pytest.mark.skip(reason="https://nvbugs/5575841")
@pytest.mark.parametrize("num_tokens", [1, 1024])
@pytest.mark.parametrize("hidden_size", [1024])
@pytest.mark.parametrize("intermediate_size", [1024, 768, 384, 192])
@ -1078,7 +1077,6 @@ class TestMoeFp4:
use_autotune=True,
use_topk_as_input=False)
@pytest.mark.skip(reason="https://nvbugs/5575841")
@pytest.mark.parametrize("num_tokens", [1, 150])
@pytest.mark.parametrize("hidden_size", [1024])
@pytest.mark.parametrize("intermediate_size", [1024])