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
[Bug] Fix AttributeError: 'FusedMoE' object has no attribute 'w13_weight_scale'. Did you mean: 'w13_weight_scale_inv' (#25519)
Signed-off-by: yewentao256 <zhyanwentao@126.com> Signed-off-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com> Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Wentao Ye committed
Sep 23, 2025 at 20:07 UTC
88d7bdbd2337917fbdd65bdfe33e6af7ecdb45cd
1 file changed
+4
-2
vllm/model_executor/warmup/deep_gemm_warmup.py
+4
-2
@@ -53,9 +53,11 @@ def _extract_data_from_fused_moe_module(
53
"""
54
assert isinstance(m, FusedMoE)
55
w13 = m.w13_weight
56
- w13_s = getattr(m, "w13_weight_scale_inv", m.w13_weight_scale)
56
+ w13_s = m.w13_weight_scale_inv if hasattr(
57
+ m, "w13_weight_scale_inv") else m.w13_weight_scale
58
w2 = m.w2_weight
58
- w2_s = getattr(m, "w2_weight_scale_inv", m.w2_weight_scale)
59
+ w2_s = m.w2_weight_scale_inv if hasattr(
60
+ m, "w2_weight_scale_inv") else m.w2_weight_scale
61
num_topk = m.top_k
62
63
assert isinstance(w13, torch.Tensor)