{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "b7672826cad31e30319487af876e608d8af7d37b": { "metrics": { "prefill.latency.mean": 0.01970032024383545, "prefill.throughput.value": 355.3241730773597, "decode.latency.mean": 2.3911510009765626, "decode.throughput.value": 53.11249684697132, "per_token.latency.mean": 0.01890306428298649, "per_token.throughput.value": 52.90147592102513 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "b7672826cad31e30319487af876e608d8af7d37b": { "metrics": { "prefill.latency.mean": 0.022671199798583987, "prefill.throughput.value": 308.76177979946215, "decode.latency.mean": 2.635069580078125, "decode.throughput.value": 48.19607078316113, "per_token.latency.mean": 0.02083684174156943, "per_token.throughput.value": 47.99191798846383 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "b7672826cad31e30319487af876e608d8af7d37b": { "metrics": { "prefill.latency.mean": 0.014078880310058595, "prefill.throughput.value": 497.19862985118783, "decode.latency.mean": 1.5618172607421874, "decode.throughput.value": 81.31553107541444, "per_token.latency.mean": 0.012350609700199173, "per_token.throughput.value": 80.9676626720601 } } } } }