Misc fix for min_p_sampling, --cuda-graph-bs (#2761)

2025-01-07 02:52:53 -08:00
parent 6d08ce2aa9
commit bdc1acf6cd
17 changed files with 135 additions and 63 deletions
--- a/benchmark/kernels/fused_moe_triton/tuning_fused_moe_triton.py
+++ b/benchmark/kernels/fused_moe_triton/tuning_fused_moe_triton.py
@@ -228,6 +228,7 @@ class BenchmarkWorker:
                hidden_size,
                topk,
                dtype_str,
+                False,
            )
        else:
            config = op_config[min(op_config.keys(), key=lambda x: abs(x - num_tokens))]