{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "bd5091df8db7cea1a9f94f797fc11487f840ade1": { "metrics": { "prefill.latency.mean": 0.020133775711059572, "prefill.throughput.value": 347.67447996129556, "decode.latency.mean": 2.4417091064453125, "decode.throughput.value": 52.0127478186331, "per_token.latency.mean": 0.019298247371266484, "per_token.throughput.value": 51.81817709979811 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "bd5091df8db7cea1a9f94f797fc11487f840ade1": { "metrics": { "prefill.latency.mean": 0.02215336036682129, "prefill.throughput.value": 315.97915097719357, "decode.latency.mean": 2.565888427734375, "decode.throughput.value": 49.49552701796091, "per_token.latency.mean": 0.020287751080961567, "per_token.throughput.value": 49.29082558285231 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "bd5091df8db7cea1a9f94f797fc11487f840ade1": { "metrics": { "prefill.latency.mean": 0.014144016265869142, "prefill.throughput.value": 494.9089331077529, "decode.latency.mean": 1.5736651611328125, "decode.throughput.value": 80.70331804802635, "per_token.latency.mean": 0.012443789655512029, "per_token.throughput.value": 80.36137122882384 } } } } }