diff --git a/config.json b/config.json index 039b816..f460f86 100644 --- a/config.json +++ b/config.json @@ -76,9 +76,9 @@ "ctx": "131072", "trust_remote": true, "valid_tp": [1, 2], - "max_num_seqs": "8", - "max_tokens": "8192", - "gpu_util": "0.92", + "max_num_seqs": "6", + "max_tokens": "4096", + "gpu_util": "0.90", "enforce_eager": false, "tool_call_parser": "qwen3_coder", "reasoning_parser": "qwen3",