{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "c35d2ccf5a5ebcab24a0b9da3cb2527b3ae565a0": { "metrics": { "prefill.latency.mean": 0.02051244831085205, "prefill.throughput.value": 341.25619204103833, "decode.latency.mean": 2.427037475585937, "decode.throughput.value": 52.32716893641684, "per_token.latency.mean": 0.019109263142262856, "per_token.throughput.value": 52.33064156138798 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "c35d2ccf5a5ebcab24a0b9da3cb2527b3ae565a0": { "metrics": { "prefill.latency.mean": 0.021654767990112303, "prefill.throughput.value": 323.25444461913617, "decode.latency.mean": 2.5708712158203126, "decode.throughput.value": 49.399596221888885, "per_token.latency.mean": 0.02024205705687756, "per_token.throughput.value": 49.402093729413444 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "c35d2ccf5a5ebcab24a0b9da3cb2527b3ae565a0": { "metrics": { "prefill.latency.mean": 0.014148575782775879, "prefill.throughput.value": 494.749443864281, "decode.latency.mean": 1.5675462036132815, "decode.throughput.value": 81.0183455564231, "per_token.latency.mean": 0.012341764037064678, "per_token.throughput.value": 81.0256943008154 } } } } }