{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "940fde8dafaecb8f17b588c5078291f1c1a420c8": { "metrics": { "prefill.latency.mean": 0.02033520030975342, "prefill.throughput.value": 344.2306883322204, "decode.latency.mean": 2.4782998046875, "decode.throughput.value": 51.244808945144555, "per_token.latency.mean": 0.01958842790173919, "per_token.throughput.value": 51.05054908011344 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "940fde8dafaecb8f17b588c5078291f1c1a420c8": { "metrics": { "prefill.latency.mean": 0.025318976402282714, "prefill.throughput.value": 276.4724722192518, "decode.latency.mean": 2.537893798828125, "decode.throughput.value": 50.04149506123636, "per_token.latency.mean": 0.020059904995643103, "per_token.throughput.value": 49.85068474737015 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "940fde8dafaecb8f17b588c5078291f1c1a420c8": { "metrics": { "prefill.latency.mean": 0.01400380802154541, "prefill.throughput.value": 499.86403621288036, "decode.latency.mean": 1.5594651489257814, "decode.throughput.value": 81.4381777543938, "per_token.latency.mean": 0.012331704181173573, "per_token.throughput.value": 81.09179277318934 } } } } }