{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "66bc4def9505fa7c7fe4aa7a248c34a026bb552b": { "metrics": { "prefill.latency.mean": 0.02079647922515869, "prefill.throughput.value": 336.59543638192855, "decode.latency.mean": 2.515798828125, "decode.throughput.value": 50.4809838450604, "per_token.latency.mean": 0.019808215719508374, "per_token.throughput.value": 50.48410286723288 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "66bc4def9505fa7c7fe4aa7a248c34a026bb552b": { "metrics": { "prefill.latency.mean": 0.022945216178894043, "prefill.throughput.value": 305.0744846082073, "decode.latency.mean": 2.6843699951171875, "decode.throughput.value": 47.31091475132352, "per_token.latency.mean": 0.02113573500115102, "per_token.throughput.value": 47.31323514159983 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "66bc4def9505fa7c7fe4aa7a248c34a026bb552b": { "metrics": { "prefill.latency.mean": 0.014140079975128174, "prefill.throughput.value": 495.0467049912529, "decode.latency.mean": 1.5638016357421876, "decode.throughput.value": 81.21234630869613, "per_token.latency.mean": 0.01231224945398766, "per_token.throughput.value": 81.21992684903915 } } } } }