{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "25245ec26dc29bcf6102e1b4ddd0dfd02e720cf5": { "metrics": { "prefill.latency.mean": 0.022692079544067385, "prefill.throughput.value": 308.47767770275067, "decode.latency.mean": 2.385437255859375, "decode.throughput.value": 53.23971514574468, "per_token.latency.mean": 0.018857490215376905, "per_token.throughput.value": 53.029326202941526 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "25245ec26dc29bcf6102e1b4ddd0dfd02e720cf5": { "metrics": { "prefill.latency.mean": 0.021704048156738282, "prefill.throughput.value": 322.52047864291, "decode.latency.mean": 2.5922081298828124, "decode.throughput.value": 48.99297958985314, "per_token.latency.mean": 0.020496448750552454, "per_token.throughput.value": 48.78893959486744 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "25245ec26dc29bcf6102e1b4ddd0dfd02e720cf5": { "metrics": { "prefill.latency.mean": 0.01405294418334961, "prefill.throughput.value": 498.1162600997043, "decode.latency.mean": 1.5584067993164064, "decode.throughput.value": 81.49348427875726, "per_token.latency.mean": 0.012323451479433083, "per_token.throughput.value": 81.14609788247432 } } } } }