From 366a0240a517511ed35c2956c01d6c5f78b6050e Mon Sep 17 00:00:00 2001 From: biondizzle Date: Sun, 17 May 2026 07:14:58 +0000 Subject: [PATCH] vllm tweaks --- docker-compose.yml | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/docker-compose.yml b/docker-compose.yml index 997732dd..30e5e48c 100644 --- a/docker-compose.yml +++ b/docker-compose.yml @@ -19,12 +19,12 @@ services: - --compilation-config #- '{"cudagraph_mode": "FULL_DECODE_ONLY", "custom_ops": ["all"], "cudagraph_capture_sizes": [1, 2, 4, 8], "max_cudagraph_capture_size": 8}' - '{"cudagraph_mode":"FULL_AND_PIECEWISE", "custom_ops":["all"]}' - - --moe-backend deep_gemm_mega_moe + #- --moe-backend=deep_gemm_mega_moe - --tokenizer-mode=deepseek_v4 #- --attention_config.use_fp4_indexer_cache=True - - --tool-call-parser deepseek_v4 + - --tool-call-parser=deepseek_v4 - --enable-auto-tool-choice - - --reasoning-parser deepseek_v4 + - --reasoning-parser=deepseek_v4 - --host=0.0.0.0 - --port=8000 deploy: