diff --git a/docker-compose.yml b/docker-compose.yml index 11cf955f..4b6217fb 100644 --- a/docker-compose.yml +++ b/docker-compose.yml @@ -15,7 +15,8 @@ services: - --trust-remote-code - --enable-expert-parallel - --tensor-parallel-size=8 - - --enforce-eager + - --compilation-config + - '{"cudagraph_mode": "FULL_DECODE_ONLY", "custom_ops": ["all"], "cudagraph_capture_sizes": [1, 2, 4, 8], "max_cudagraph_capture_size": 8}' - --tokenizer-mode=deepseek_v4 - --host=0.0.0.0 - --port=8000