examples/bursty.json

{
  "model": "Qwen2.5-3B",
  "accelerator": "L4",
  "scheduler": "chunked_slo",
  "quantization": "int8",
  "arrival_process": "bursty",
  "request_rate_rps": 3.0,
  "duration_s": 60,
  "prompt_tokens_mean": 1024,
  "prompt_tokens_cv": 0.8,
  "output_tokens_mean": 96,
  "output_tokens_cv": 0.6,
  "max_batch_size": 16,
  "max_batch_tokens": 8192,
  "chunk_size": 512,
  "slo_ttft_ms": 750,
  "slo_e2e_ms": 8000,
  "slo_attainment_target": 0.95,
  "seed": 11
}