{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "f745e7d3f902601686b83c7cce2660c2a94509f0": { "metrics": { "prefill.latency.mean": 0.021567856788635254, "prefill.throughput.value": 324.5570511989169, "decode.latency.mean": 2.60877099609375, "decode.throughput.value": 48.68192730989565, "per_token.latency.mean": 0.02054022249086635, "per_token.throughput.value": 48.684964364172366 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "f745e7d3f902601686b83c7cce2660c2a94509f0": { "metrics": { "prefill.latency.mean": 0.02229771137237549, "prefill.throughput.value": 313.9335639922338, "decode.latency.mean": 2.6286884765625, "decode.throughput.value": 48.313066052648495, "per_token.latency.mean": 0.020697333951634683, "per_token.throughput.value": 48.31540150711148 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "f745e7d3f902601686b83c7cce2660c2a94509f0": { "metrics": { "prefill.latency.mean": 0.01422929573059082, "prefill.throughput.value": 491.9428292540906, "decode.latency.mean": 1.564611572265625, "decode.throughput.value": 81.17030594123659, "per_token.latency.mean": 0.01231867967064925, "per_token.throughput.value": 81.17753093155116 } } } } }