{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "0ed3ffcb4461a244b87781a24e5ebd0a78f98142": { "metrics": { "prefill.latency.mean": 0.01963271999359131, "prefill.throughput.value": 356.54764099345397, "decode.latency.mean": 2.3445885009765624, "decode.throughput.value": 54.16728775522966, "per_token.latency.mean": 0.01853426648317118, "per_token.throughput.value": 53.954117952711215 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "0ed3ffcb4461a244b87781a24e5ebd0a78f98142": { "metrics": { "prefill.latency.mean": 0.021912784576416017, "prefill.throughput.value": 319.44821871401325, "decode.latency.mean": 2.5411068115234374, "decode.throughput.value": 49.97822186146568, "per_token.latency.mean": 0.020091584269708324, "per_token.throughput.value": 49.77208300630029 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "0ed3ffcb4461a244b87781a24e5ebd0a78f98142": { "metrics": { "prefill.latency.mean": 0.014057071685791015, "prefill.throughput.value": 497.9700008982417, "decode.latency.mean": 1.5599285888671874, "decode.throughput.value": 81.4139832466477, "per_token.latency.mean": 0.012335715033791281, "per_token.throughput.value": 81.06542646783713 } } } } }