[
  {
    "model_name": "Baseline",
    "batch_size": 32,
    "seq_len": 81,
    "hidden_size": 512,
    "param_count": 27296770,
    "latency_mean_ms": 12.367569541931152,
    "latency_std_ms": 0.03661433936324098,
    "throughput_sps": 2587.4121743570417,
    "gpu_memory_mb": 174.16943359375
  },
  {
    "model_name": "Tiered",
    "batch_size": 32,
    "seq_len": 81,
    "hidden_size": 512,
    "param_count": 27296770,
    "latency_mean_ms": 11.73528323173523,
    "latency_std_ms": 0.06700267067328709,
    "throughput_sps": 2726.81957206314,
    "gpu_memory_mb": 193.54443359375
  }
]