From c6850ac30cbea6e751575beb150818e757c710b2 Mon Sep 17 00:00:00 2001 From: Michael <13900043+michaelzhang-ai@users.noreply.github.com> Date: Tue, 3 Mar 2026 22:01:08 -0800 Subject: [PATCH] [AMD] Fix Qwen3-Coder-Next: Add missing k_scale/v_scale args to extend_attention_fwd in aiter_backend (#19736) --- python/sglang/srt/layers/attention/aiter_backend.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/python/sglang/srt/layers/attention/aiter_backend.py b/python/sglang/srt/layers/attention/aiter_backend.py index 5d29b16d7..dd48ae19a 100755 --- a/python/sglang/srt/layers/attention/aiter_backend.py +++ b/python/sglang/srt/layers/attention/aiter_backend.py @@ -1765,6 +1765,8 @@ class AiterAttnBackend(AttentionBackend): True, # causal self.forward_metadata.mask_indptr, self.forward_metadata.max_extend_len, + 1.0, # k_scale + 1.0, # v_scale layer.scaling, logit_cap=layer.logit_cap, )