1

2026-03-30 22:12:05 +08:00
parent f82de29267
commit 6f18a26bd9
1 changed files with 6 additions and 6 deletions
@@ -73,13 +73,13 @@
        "local_path": "Qwen3.5-35B-A3B-GPTQ-Int4",
        "dtype": "float16",
        "quantization": "gptq",
-        "ctx": "65536",
+        "ctx": "32768",
        "trust_remote": true,
-        "valid_tp": [2],
+        "valid_tp": [1],
-        "max_num_seqs": "2",
+        "max_num_seqs": "1",
-        "max_tokens": "2048",
+        "max_tokens": "1024",
-        "gpu_util": "0.90",
+        "gpu_util": "0.88",
-        "enforce_eager": false,
+        "enforce_eager": true,
        "tool_call_parser": "qwen3_xml",
        "reasoning_parser": "qwen3",
        "enable_auto_tool_choice": true,