From 454676811ec7fdf4f0dea43607a5fdcc02c38ddb Mon Sep 17 00:00:00 2001 From: YAMY <74099316+YAMY1234@users.noreply.github.com> Date: Wed, 11 Feb 2026 21:31:27 -0800 Subject: [PATCH] [Flashinfer Autotune] Fix FlashInfer FP4 MoE autotuning crash by removing incorrect flatten on hidden_states_scale (#18500) --- python/sglang/srt/layers/moe/fused_moe_triton/layer.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/python/sglang/srt/layers/moe/fused_moe_triton/layer.py b/python/sglang/srt/layers/moe/fused_moe_triton/layer.py index ae70a66e2..1609b6dbb 100644 --- a/python/sglang/srt/layers/moe/fused_moe_triton/layer.py +++ b/python/sglang/srt/layers/moe/fused_moe_triton/layer.py @@ -1329,7 +1329,7 @@ class FlashInferFP4MoE(FusedMoE): routing_logits=router_logits, routing_bias=correction_bias, hidden_states=hs_fp4, - hidden_states_scale=hs_scale_linear.view(torch.float8_e4m3fn).flatten(), + hidden_states_scale=hs_scale_linear.view(torch.float8_e4m3fn), gemm1_weights=self.gemm1_weights_fp4_shuffled.data, gemm1_weights_scale=self.gemm1_scales_fp4_shuffled.data.view( torch.float8_e4m3fn