GuideLLM Benchmark Testing Best Practice

conda create -n guidellm-bench python=3.11 -y
conda activate guidellm-bench

git clone https://github.com/vllm-project/guidellm.git
cd guidellm
pip install guidellm

guidellm --help

docker pull ghcr.io/llm-d/llm-d-inference-sim:v0.4.0

docker run --rm --publish 8000:8000 \
ghcr.io/llm-d/llm-d-inference-sim:v0.4.0  \
--port 8000 \
--model "Qwen/Qwen2.5-1.5B-Instruct"  \
--lora-modules '{"name":"tweet-summary-0"}' '{"name":"tweet-summary-1"}'

curl --request GET 'http://localhost:8000/v1/models'

curl --request POST 'http://localhost:8000/v1/chat/completions' \
--header 'Content-Type: application/json' \
--data-raw '{
    "model": "tweet-summary-0",
    "stream": false,
    "messages": [{"role": "user", "content": "Say this is a test!"}]
}'

curl --request POST 'http://localhost:8000/v1/completions' \
--header 'Content-Type: application/json' \
--data-raw '{
    "model": "tweet-summary-0",
    "stream": false,
    "prompt": "Say this is a test!",
    "max_tokens": 128
}'

ls ./Qwen2.5-1.5B-Instruct
merges.txt              tokenizer.json          tokenizer_config.json   vocab.json

guidellm run \
  --backend kind=openai_http,target=http://localhost:8000/,model=tweet-summary-0 \
  --tokenizer kind=huggingface_auto,model=${local_path}/Qwen2.5-1.5B-Instruct \
  --profile kind=sweep \
  --constraint kind=max_duration,seconds=10 \
  --constraint kind=max_requests,count=10 \
  --data kind=synthetic_text,prompt_tokens=128,output_tokens=56