diff --git a/python/sglang/srt/layers/k3_ar_fusion.py b/python/sglang/srt/layers/k3_ar_fusion.py index b91774a5c..d5e0a116e 100644 --- a/python/sglang/srt/layers/k3_ar_fusion.py +++ b/python/sglang/srt/layers/k3_ar_fusion.py @@ -371,6 +371,8 @@ def finalize_all_reduce_push_norm( state = _get_state() assert state is not None + if expert_weights.dtype != torch.bfloat16: + expert_weights = expert_weights.to(torch.bfloat16) return mod.finalize_all_reduce_push_norm( state.world_size, out,