{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "9ba9369a2557e53a01378199a9839ec6e82d8bc7": { "metrics": { "prefill.latency.mean": 0.019696576118469238, "prefill.throughput.value": 355.3917167073614, "decode.latency.mean": 2.428328125, "decode.throughput.value": 52.29935719663091, "per_token.latency.mean": 0.01919638968456404, "per_token.throughput.value": 52.09312878265373 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "9ba9369a2557e53a01378199a9839ec6e82d8bc7": { "metrics": { "prefill.latency.mean": 0.022163279533386232, "prefill.throughput.value": 315.8377346391976, "decode.latency.mean": 2.604711669921875, "decode.throughput.value": 48.75779590752523, "per_token.latency.mean": 0.020594631338307978, "per_token.throughput.value": 48.556343814705954 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "9ba9369a2557e53a01378199a9839ec6e82d8bc7": { "metrics": { "prefill.latency.mean": 0.014070287704467773, "prefill.throughput.value": 497.50226484546386, "decode.latency.mean": 1.5731578369140626, "decode.throughput.value": 80.72934388397142, "per_token.latency.mean": 0.012440422197575626, "per_token.throughput.value": 80.38312399034808 } } } } }