API url for accessing a new version of server (V0.2.4)

[lililolo0927]

Hello, I just tried new version of docker image of KTransformers, V0.2.4 and when I ran the cmd as follows ,

PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True python3 ktransformers/server/main.py \
--gguf_path /models/gguf/DeepSeek-V3-0324 \
--model_path deepseek-ai/DeepSeek-V3 \
--model_name DeepSeek-V3 \
--cpu_infer 17 \
--optimize_config_path ktransformers/optimize/optimize_rules/DeepSeek-V3-Chat-multi-gpu.yaml \
--no-use_cuda_graph \
--backend_type balance_serve \
--port 10002

then it shows the following output,

2025-04-11 11:21:43,392 - INFO - flashinfer.jit: Prebuilt kernels not found, using JIT backend
found flashinfer
sched_rpc started with PID: 34
set start method
tokenizer_config.json: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3.59k/3.59k [00:00<00:00, 32.5MB/s]
tokenizer.json: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7.85M/7.85M [00:34<00:00, 230kB/s]
Connected to server at tcp://localhost:38709
2025-04-11 11:22:23,650 - INFO - flashinfer.jit: Prebuilt kernels not found, using JIT backend
found flashinfer
start method already set to spawn
Connected to server at tcp://localhost:38709
Getting inference context from sched_client.

and then it cannot connect to the server just deployed showing the following output,

$ curl -X POST http://localhost:10002/v1/chat/completions   -H "accept: application/json"   -H "Content-Type: applicatio                                                                                                                                 n/json"   -d '{
    "messages": [
      {"role": "user", "content": "hello"}
    ],
    "model": "DeepSeek-V3",
    "temperature": 0.3,
    "top_p": 1.0,
    "stream": true
  }'
curl: (7) Failed to connect to localhost port 10002 after 0 ms: Couldn't connect to server

I tried with the port 38709 but returns same result.

Can anyone help me? (The previous version v0.2.3 is working well, so i dont think the port of my server is pb)