{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "74b92c62560b7ade42d35a49f9063adc8b805c4a": { "metrics": { "prefill.latency.mean": 0.019320960044860842, "prefill.throughput.value": 362.30083721237867, "decode.latency.mean": 2.3275152587890626, "decode.throughput.value": 54.564626169658, "per_token.latency.mean": 0.01840009012335374, "per_token.throughput.value": 54.34755989215407 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "74b92c62560b7ade42d35a49f9063adc8b805c4a": { "metrics": { "prefill.latency.mean": 0.022983008384704587, "prefill.throughput.value": 304.57283410550235, "decode.latency.mean": 2.6945513916015624, "decode.throughput.value": 47.132149862064765, "per_token.latency.mean": 0.021307865519768633, "per_token.throughput.value": 46.93102643585946 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "74b92c62560b7ade42d35a49f9063adc8b805c4a": { "metrics": { "prefill.latency.mean": 0.014070799827575683, "prefill.throughput.value": 497.4841576725109, "decode.latency.mean": 1.559213195800781, "decode.throughput.value": 81.4513373424699, "per_token.latency.mean": 0.012329813972292209, "per_token.throughput.value": 81.10422446333894 } } } } }