1 parent 721f45d commit 6a1d0afCopy full SHA for 6a1d0af
2 files changed
tests/benchmarks/configs/musa.yaml
@@ -0,0 +1,23 @@
1
+serve:
2
+ - name: serve_smoke_dummy
3
+ model: qwen3
4
+ case: 4b_tp2_eager_musa
5
+ server_parameters:
6
+ served_model_name: smoke-model
7
+ tensor_parallel_size: 1
8
+ max_model_len: 1024
9
+ gpu_memory_utilization: 0.7
10
+ max_num_seqs: 4
11
+ enforce_eager: true
12
+ disable_custom_all_reduce: true
13
+ load_format: dummy
14
+ client_parameters:
15
+ model: smoke-model
16
+ backend: openai-chat
17
+ endpoint: /v1/chat/completions
18
+ dataset_name: random
19
+ random_input_len: 32
20
+ random_output_len: 4
21
+ num_prompts: 5
22
+ percentile_metrics: ttft,tpot,itl,e2el
23
+ goodput: e2el:10000
0 commit comments