benchmark/llm_text_generation/chat/pegasus/A100/queue_2gpu.yaml

- command:
    - "python scripts/benchmark_one_model.py --model {{ model }} --result-root results/joule --gpu-ids {{ gpu }} --backends vllm --server-images mlenergy/vllm:v0.5.4-openai --request-rate inf --power-limits 400 --max-num-seqs 320 256 192 128 63 32 --data-dup-factor 1"
  model:
    - google/gemma-2-27b-it