benchmarking/tinystories_2_5m_eval_metrics.json

{
  "model": "2.5M",
  "num_samples": 100,
  "temperature": 0.8,
  "top_k": 50,
  "metrics": {
    "ttft_ms": 4.13,
    "tps": 538.3,
    "total_latency_ms": 68.7,
    "tokens_generated": 37.1,
    "perplexity": 3.83,
    "avg_token_prob": 0.5382,
    "repetition_rate": 0.0682,
    "coherence_length": 27.3,
    "vocab_diversity": 0.8371
  },
  "evaluated_at": "2026-06-21 09:19:22"
}