{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "970a16ec7f693104085fd826523e5c6ce64f2040": { "metrics": { "prefill.latency.mean": 0.020231648445129397, "prefill.throughput.value": 345.9925679800547, "decode.latency.mean": 2.4315745849609374, "decode.throughput.value": 52.2295309325419, "per_token.latency.mean": 0.019144986190195157, "per_token.throughput.value": 52.232996674196414 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "970a16ec7f693104085fd826523e5c6ce64f2040": { "metrics": { "prefill.latency.mean": 0.021830528259277344, "prefill.throughput.value": 320.65188331048296, "decode.latency.mean": 2.566541015625, "decode.throughput.value": 49.48294191553107, "per_token.latency.mean": 0.020207986861702026, "per_token.throughput.value": 49.48538450879489 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "970a16ec7f693104085fd826523e5c6ce64f2040": { "metrics": { "prefill.latency.mean": 0.014175551891326904, "prefill.throughput.value": 493.8079345103201, "decode.latency.mean": 1.565439697265625, "decode.throughput.value": 81.12736646568541, "per_token.latency.mean": 0.012325130297443058, "per_token.throughput.value": 81.13504489339618 } } } } }