{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "746104ba6f0514159d58dc2fb09c887d0e9d4863": { "metrics": { "prefill.latency.mean": 0.02057169532775879, "prefill.throughput.value": 340.2733653436148, "decode.latency.mean": 2.56627685546875, "decode.throughput.value": 49.48803545079804, "per_token.latency.mean": 0.02020562039773295, "per_token.throughput.value": 49.491180192230026 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "746104ba6f0514159d58dc2fb09c887d0e9d4863": { "metrics": { "prefill.latency.mean": 0.02242950439453125, "prefill.throughput.value": 312.08892879981494, "decode.latency.mean": 2.6397170410156248, "decode.throughput.value": 48.111217235290134, "per_token.latency.mean": 0.020784152241203724, "per_token.throughput.value": 48.11358136693886 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "746104ba6f0514159d58dc2fb09c887d0e9d4863": { "metrics": { "prefill.latency.mean": 0.01416761589050293, "prefill.throughput.value": 494.08454140067107, "decode.latency.mean": 1.56476025390625, "decode.throughput.value": 81.16259323622172, "per_token.latency.mean": 0.012319836729154812, "per_token.throughput.value": 81.16990687331973 } } } } }