x

2026-04-23 17:19:46 +08:00
parent 6a3403a530
commit 6e4e952e20
1 changed files with 2 additions and 2 deletions
@@ -71,14 +71,14 @@
        "quantization": "fp8",
        "ctx": "262144",
        "max_tokens": "65536",
-        "max_num_batched_tokens": 16384,
+        "max_num_batched_tokens": 32768,
        "trust_remote": true,
        "enforce_eager": false,
        "valid_tp": [
          1,
          2
        ],
-        "max_num_seqs": 32,
+        "max_num_seqs": 256,
        "gpu_util": "0.92",
        "kv_cache_dtype": "fp8",
        "enable_prefix_caching": true,