{
    "google/gemma-2b": {
        "backend.cache_implementation=null,backend.torch_compile=False": {
            "5c1027bf09717f664b579e01cbb8ec3ef5aeb140": {
                "metrics": {
                    "prefill.latency.mean": 0.02060132789611816,
                    "prefill.throughput.value": 339.78392243924173,
                    "decode.latency.mean": 2.452270751953125,
                    "decode.throughput.value": 51.78873495059635,
                    "per_token.latency.mean": 0.019307854607349306,
                    "per_token.throughput.value": 51.79239332055887
                }
            }
        },
        "backend.cache_implementation=static,backend.torch_compile=False": {
            "5c1027bf09717f664b579e01cbb8ec3ef5aeb140": {
                "metrics": {
                    "prefill.latency.mean": 0.022531567573547365,
                    "prefill.throughput.value": 310.67523274404476,
                    "decode.latency.mean": 2.600800537109375,
                    "decode.throughput.value": 48.83111879896504,
                    "per_token.latency.mean": 0.0204777060080701,
                    "per_token.throughput.value": 48.83359491565648
                }
            }
        },
        "backend.cache_implementation=static,backend.torch_compile=True": {
            "5c1027bf09717f664b579e01cbb8ec3ef5aeb140": {
                "metrics": {
                    "prefill.latency.mean": 0.014148736000061035,
                    "prefill.throughput.value": 494.7438414265277,
                    "decode.latency.mean": 1.5659439086914064,
                    "decode.throughput.value": 81.10124462001232,
                    "per_token.latency.mean": 0.01232914539397232,
                    "per_token.throughput.value": 81.10862253996103
                }
            }
        }
    }
}