2024-06-12/summary.json

{
    "google/gemma-2b": {
        "backend.cache_implementation=null,backend.torch_compile=False": {
            "35a6d9d6483d4d4d7cd817ed4ecfd5f86e1f9a23": {
                "metrics": {
                    "prefill.latency.mean": 0.020440783500671387,
                    "prefill.throughput.value": 342.45262662119,
                    "decode.latency.mean": 2.45060498046875,
                    "decode.throughput.value": 51.82393776728044,
                    "per_token.latency.mean": 0.019376820123242766,
                    "per_token.throughput.value": 51.608055069907266
                }
            }
        },
        "backend.cache_implementation=static,backend.torch_compile=False": {
            "35a6d9d6483d4d4d7cd817ed4ecfd5f86e1f9a23": {
                "metrics": {
                    "prefill.latency.mean": 0.021873536109924315,
                    "prefill.throughput.value": 320.0214160537128,
                    "decode.latency.mean": 2.4960036621093753,
                    "decode.throughput.value": 50.88133560375956,
                    "per_token.latency.mean": 0.019735539817056166,
                    "per_token.throughput.value": 50.67001000579492
                }
            }
        },
        "backend.cache_implementation=static,backend.torch_compile=True": {
            "35a6d9d6483d4d4d7cd817ed4ecfd5f86e1f9a23": {
                "metrics": {
                    "prefill.latency.mean": 0.014046736240386963,
                    "prefill.throughput.value": 498.33640215110654,
                    "decode.latency.mean": 1.5604882202148436,
                    "decode.throughput.value": 81.38478609118562,
                    "per_token.latency.mean": 0.012340118773841105,
                    "per_token.throughput.value": 81.03649716239565
                }
            }
        }
    }
}