{
    "google/gemma-2b": {
        "backend.cache_implementation=null,backend.torch_compile=False": {
            "9230d78e76611cfa38c845213021aeb185362d10": {
                "metrics": {
                    "prefill.latency.mean": 0.020968351364135742,
                    "prefill.throughput.value": 333.8364508700859,
                    "decode.latency.mean": 2.44393505859375,
                    "decode.throughput.value": 51.96537426533597,
                    "per_token.latency.mean": 0.019242322673947796,
                    "per_token.throughput.value": 51.968778247020104
                }
            }
        },
        "backend.cache_implementation=static,backend.torch_compile=False": {
            "9230d78e76611cfa38c845213021aeb185362d10": {
                "metrics": {
                    "prefill.latency.mean": 0.021651376724243165,
                    "prefill.throughput.value": 323.30507612303757,
                    "decode.latency.mean": 2.56781884765625,
                    "decode.throughput.value": 49.45831755846715,
                    "per_token.latency.mean": 0.020218033437653788,
                    "per_token.throughput.value": 49.46079464571533
                }
            }
        },
        "backend.cache_implementation=static,backend.torch_compile=True": {
            "9230d78e76611cfa38c845213021aeb185362d10": {
                "metrics": {
                    "prefill.latency.mean": 0.014129024028778078,
                    "prefill.throughput.value": 495.43407851401196,
                    "decode.latency.mean": 1.5649970092773438,
                    "decode.throughput.value": 81.15031482305757,
                    "per_token.latency.mean": 0.012321654931766781,
                    "per_token.throughput.value": 81.15792931531249
                }
            }
        }
    }
}