A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
fix(tests): Resolve late binding of loop variable in assert message lambda (#26249)
Signed-off-by: lyd1992 <liuyudong@iscas.ac.cn> Signed-off-by: ihb2032 <1355790728@qq.com
ihb2032 committed
Oct 6, 2025 at 00:18 UTC
5f317530ec23444da83cc15c0d3a51c4c2239d5a
1 file changed
+6
-6
tests/kernels/mamba/test_mamba_ssm_ssd.py
+6
-6
@@ -562,15 +562,15 @@ def test_mamba_chunk_scan_cont_batch_prefill_chunking(chunk_size, seqlens):
562
Y_ref_seq[: chunked_seqlens[i], ...],
563
atol=atol,
564
rtol=rtol,
565
- msg=lambda x: f"seq{i} output part1 " + x,
566
- ) # noqa: B023
565
+ msg=lambda x, i=i: f"seq{i} output part1 " + x,
566
+ )
567
torch.testing.assert_close(
568
Y_seq[chunked_seqlens[i] :, ...],
569
Y_ref_seq[chunked_seqlens[i] :, ...],
570
atol=atol,
571
rtol=rtol,
572
- msg=lambda x: f"seq{i} output part2 " + x,
573
- ) # noqa: B023
572
+ msg=lambda x, i=i: f"seq{i} output part2 " + x,
573
+ )
574
575
state_seq = state_chunked[i]
576
state_seq_ref = state_ref[i]
@@ -579,5 +579,5 @@ def test_mamba_chunk_scan_cont_batch_prefill_chunking(chunk_size, seqlens):
579
state_seq_ref,
580
atol=atol,
581
rtol=rtol,
582
- msg=lambda x: f"seq{i} state " + x,
583
- ) # noqa: B023
582
+ msg=lambda x, i=i: f"seq{i} state " + x,
583
+ )