{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "65bb28444849976f853063edb958b3ef3dd59d12": { "metrics": { "prefill.latency.mean": 0.02127622413635254, "prefill.throughput.value": 329.00574627994285, "decode.latency.mean": 2.5274194335937503, "decode.throughput.value": 50.248881650568805, "per_token.latency.mean": 0.01989969438267505, "per_token.throughput.value": 50.25202803469252 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "65bb28444849976f853063edb958b3ef3dd59d12": { "metrics": { "prefill.latency.mean": 0.021814751625061038, "prefill.throughput.value": 320.88378177812115, "decode.latency.mean": 2.6390048828125, "decode.throughput.value": 48.1242004617478, "per_token.latency.mean": 0.020778556455777385, "per_token.throughput.value": 48.12653863266591 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "65bb28444849976f853063edb958b3ef3dd59d12": { "metrics": { "prefill.latency.mean": 0.014200816154479979, "prefill.throughput.value": 492.9294150316626, "decode.latency.mean": 1.5642773437500002, "decode.throughput.value": 81.18764904920651, "per_token.latency.mean": 0.012315990654502328, "per_token.throughput.value": 81.19525485628981 } } } } }