cjw/rocm_vllm_nightly
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

x

Browse Source
This commit is contained in:
chenjw28
2026-04-23 17:19:46 +08:00
parent 6a3403a530
commit 6e4e952e20
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
config.json
+2 -2
View File
@@ -71,14 +71,14 @@
"quantization": "fp8", "quantization": "fp8",
"ctx": "262144", "ctx": "262144",
"max_tokens": "65536", "max_tokens": "65536",
"max_num_batched_tokens": 16384, "max_num_batched_tokens": 32768,
"trust_remote": true, "trust_remote": true,
"enforce_eager": false, "enforce_eager": false,
"valid_tp": [ "valid_tp": [
1, 1,
2 2
], ],
"max_num_seqs": 32, "max_num_seqs": 256,
"gpu_util": "0.92", "gpu_util": "0.92",
"kv_cache_dtype": "fp8", "kv_cache_dtype": "fp8",
"enable_prefix_caching": true, "enable_prefix_caching": true,