update

Signed-off-by: ttanzhiqiang <389825161@qq.com>

Author: ttanzhiqiang

examples/run_dp_attention_etp16.sh

@@ -18,6 +18,6 @@ nohup python -m vllm.entrypoints.openai.api_server --model=/mnt/deepseek/DeepSee
     --max-num-batched-tokens 32768 \
     --block-size 128 \
     --no-enable-prefix-caching \
-    --additional-config '{"torchair_graph_batch_sizes":[24],"expert_tensor_parallel_size":16,"use_cached_npu_graph":true,"ascend_scheduler_config":{},"enable_graph_mode":true}' \
+    --additional-config '{"torchair_graph_config":{"enabled":true,"use_cached_graph":true,"graph_batch_sizes":[24]},"ascend_scheduler_config":{"enabled":true},"expert_tensor_parallel_size":16}' \
     --gpu-memory-utilization 0.96 &> run.log &
 disown