File tree Expand file tree Collapse file tree 1 file changed +25
-3
lines changed
Expand file tree Collapse file tree 1 file changed +25
-3
lines changed Original file line number Diff line number Diff line change @@ -57,7 +57,29 @@ python benchmark_serving.py \
5757 --backend < backend> # By default <backend> is vllm
5858 --dataset-path ShareGPT_V3_unfiltered_cleaned_split.json \
5959 --dataset-name sharegpt \
60- --model /data/mtt/models_convert/< convert_model_dir> \
61- --trust-remote-code
62- --num-prompts < num_prompts> # By default <num_prompts> is 1000
60+ --model /data/mtt/models_convert/< convert_model_dir> \
61+ --trust-remote-code \
62+ --num-prompts < num_prompts> # By default <num_prompts> is 1000
6363```
64+
65+ ## 最大并发数、指定输入输出长度
66+ 如需指定最大并发数和输入输出长度,请将本目录下benchmark_serving.py拷贝到/home/workspace/vllm_mtt/benchmarks下
67+ ``` shell
68+ python benchmark_serving.py \
69+ --backend vllm \
70+ --dataset-name random \
71+ --model /data/mtt/models_convert/< convert_model_dir> \
72+ --trust-remote-code \
73+ --num-prompts 1000 \
74+ --random-input-len 1024 \
75+ --random-output-len 128 \
76+ --max-concurrency 128
77+ ```
78+ ### 参数说明
79+
80+ | 参数| 含义|
81+ | ---| ---|
82+ | --max-concurrency| 最大并发数|
83+ | --random-input-len| 使用随机数时输入长度|
84+ | --random-output-len| 输出长度|
85+ | --dataset-name| 指定输入数据为随机数生成|
You can’t perform that action at this time.
0 commit comments