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, 23:45 UTC
[Bugfix] DeepGemm utils : Fix hardcoded type-cast (#21517)
Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com> Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Varun Sundar Rabindranath committed
Jul 25, 2025 at 08:47 UTC
2212cd6cfb62196d1b26b0a6e4dcb10a1ea5f7d8
1 file changed
+1
-1
vllm/model_executor/layers/fused_moe/deep_gemm_utils.py
+1
-1
@@ -52,7 +52,7 @@ def compute_aligned_M(M: int, num_topk: int, local_num_experts: int,
52
@triton.jit
53
def apply_expert_map(expert_id, expert_map):
54
if expert_id != -1:
55
- expert_id = tl.load(expert_map + expert_id).to(tl.int64)
55
+ expert_id = tl.load(expert_map + expert_id).to(expert_id.dtype)
56
return expert_id
57
58