Skip to content

Commit 1a2118b

Browse files
authored
[https://nvbugs/5702793][fix] Fix uncontiguous tensor view (#9576)
Signed-off-by: shuyix <219646547+shuyixiong@users.noreply.github.com>
1 parent ad46d19 commit 1a2118b

File tree

2 files changed

+4
-4
lines changed

2 files changed

+4
-4
lines changed

tensorrt_llm/_torch/modules/fused_moe/quantization.py

Lines changed: 4 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -480,10 +480,12 @@ def load_expert_w3_w1_weight(self,
480480

481481
dst_w3_weight, dst_w1_weight = dst_w3_w1_weight.chunk(2, dim=0)
482482
if w1_weight is not None:
483-
dst_w1_weight.copy_(w1_weight_shard.view(dst_w3_w1_weight.dtype),
483+
dst_w1_weight.copy_(w1_weight_shard.contiguous().view(
484+
dst_w3_w1_weight.dtype),
484485
non_blocking=True)
485486
if w3_weight is not None:
486-
dst_w3_weight.copy_(w3_weight_shard.view(dst_w3_w1_weight.dtype),
487+
dst_w3_weight.copy_(w3_weight_shard.contiguous().view(
488+
dst_w3_w1_weight.dtype),
487489
non_blocking=True)
488490

489491
# Helper function

tests/integration/test_lists/waives.txt

Lines changed: 0 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -411,8 +411,6 @@ accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_bfloat16_4gpus[tp4-mt
411411
accuracy/test_llm_api_pytorch.py::TestDeepSeekR1::test_nvfp4_multi_gpus[throughput_tp8] SKIP (https://nvbugs/5698897)
412412
unittest/_torch/modules/tests_lora_modules/test_lora_attention_pytorch_flow_vs_trt.py::TestLoraAttentionPytorchFlowVsTRT::test_lora_attention SKIP (https://nvbugs/5701421)
413413
unittest/llmapi/test_llm_pytorch.py::test_embedding_bias_with_torch_sampler_strategies SKIP (https://nvbugs/5702791)
414-
accuracy/test_llm_api_pytorch.py::TestLlama4ScoutInstruct::test_fp4[tp4-cuda_graph=True] SKIP (https://nvbugs/5702793)
415-
accuracy/test_llm_api_pytorch.py::TestLlama4ScoutInstruct::test_fp4_chunked_prefill[tp4ep4-cuda_graph=True] SKIP (https://nvbugs/5702793)
416414
accuracy/test_llm_api_pytorch.py::TestQwen3NextInstruct::test_nvfp4[no_cuda_graph_overlap-cutlass] SKIP (https://nvbugs/5702795)
417415
accuracy/test_llm_api_pytorch.py::TestQwen3NextInstruct::test_nvfp4[tp1-cutlass] SKIP (https://nvbugs/5702795)
418416
accuracy/test_llm_api_pytorch.py::TestQwen3NextInstruct::test_nvfp4[tp4ep1-cutlass] SKIP (https://nvbugs/5702795)

0 commit comments

Comments
 (0)