Fix noaux_tc cuda Error 700 in CUDAGraph (#4174)

2025-10-08 10:00:29 +08:00 · 2025-09-23 18:41:33 +08:00
parent c96a535a5d
commit 1a6283424e
12 changed files with 334 additions and 148 deletions
--- a/fastdeploy/model_executor/layers/moe/ep.py
+++ b/fastdeploy/model_executor/layers/moe/ep.py
@@ -369,6 +369,7 @@ class EPRunner:
                    layer.top_k,
                    layer.routed_scaling_factor,
                    layer.gate_correction_bias,
+                    getattr(layer, "renormalize", True),
                )
            else:
                topk_idx, topk_weights = fastdeploy.model_executor.ops.gpu.moe_topk_select(