{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "3345ae733b6f4aeb7204a0f3e646a3cdbaad0023": { "metrics": { "prefill.latency.mean": 0.02310329532623291, "prefill.throughput.value": 302.9870804643079, "decode.latency.mean": 2.4337664794921876, "decode.throughput.value": 52.18249206328905, "per_token.latency.mean": 0.019237726136158578, "per_token.throughput.value": 51.98119533058712 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "3345ae733b6f4aeb7204a0f3e646a3cdbaad0023": { "metrics": { "prefill.latency.mean": 0.022356799125671387, "prefill.throughput.value": 313.10385537087865, "decode.latency.mean": 2.5555784912109374, "decode.throughput.value": 49.69520616829977, "per_token.latency.mean": 0.020204191924083844, "per_token.throughput.value": 49.494679309989024 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "3345ae733b6f4aeb7204a0f3e646a3cdbaad0023": { "metrics": { "prefill.latency.mean": 0.014050096035003661, "prefill.throughput.value": 498.21723513921705, "decode.latency.mean": 1.5572908325195312, "decode.throughput.value": 81.55188314730364, "per_token.latency.mean": 0.012314810176140707, "per_token.throughput.value": 81.2030381058936 } } } } }