{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "0f67ba1d741d65b07d549daf4ee157609ce4f9c1": { "metrics": { "prefill.latency.mean": 0.01959430408477783, "prefill.throughput.value": 357.2466758560754, "decode.latency.mean": 2.4298411865234373, "decode.throughput.value": 52.266790399461776, "per_token.latency.mean": 0.01920895289243917, "per_token.throughput.value": 52.05905837759693 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "0f67ba1d741d65b07d549daf4ee157609ce4f9c1": { "metrics": { "prefill.latency.mean": 0.021989712715148928, "prefill.throughput.value": 318.3306708312579, "decode.latency.mean": 2.5694271240234374, "decode.throughput.value": 49.42736021293809, "per_token.latency.mean": 0.020312974703641747, "per_token.throughput.value": 49.22961873332704 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "0f67ba1d741d65b07d549daf4ee157609ce4f9c1": { "metrics": { "prefill.latency.mean": 0.014056335926055909, "prefill.throughput.value": 497.9960664588458, "decode.latency.mean": 1.5611361694335937, "decode.throughput.value": 81.35100735387978, "per_token.latency.mean": 0.012345267096055825, "per_token.throughput.value": 81.002702672953 } } } } }