{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "1c68f2cafb4ca54562f74b66d1085b68dd6682f5": { "metrics": { "prefill.latency.mean": 0.01942091178894043, "prefill.throughput.value": 360.43621824111625, "decode.latency.mean": 2.3815675048828124, "decode.throughput.value": 53.32622306091179, "per_token.latency.mean": 0.01882430539489264, "per_token.throughput.value": 53.12281005977078 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "1c68f2cafb4ca54562f74b66d1085b68dd6682f5": { "metrics": { "prefill.latency.mean": 0.023109519958496094, "prefill.throughput.value": 302.9054698051608, "decode.latency.mean": 2.6399566650390627, "decode.throughput.value": 48.10685026836258, "per_token.latency.mean": 0.020877902826301663, "per_token.throughput.value": 47.897531103565406 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "1c68f2cafb4ca54562f74b66d1085b68dd6682f5": { "metrics": { "prefill.latency.mean": 0.014141839981079103, "prefill.throughput.value": 494.98509453971775, "decode.latency.mean": 1.576483947753906, "decode.throughput.value": 80.55901880951158, "per_token.latency.mean": 0.01246668597737791, "per_token.throughput.value": 80.21377949317112 } } } } }