{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "0a7af19f4dc868bafc82f35eb7e8d13bac87a594": { "metrics": { "prefill.latency.mean": 0.020197423934936523, "prefill.throughput.value": 346.57885196397444, "decode.latency.mean": 2.43524609375, "decode.throughput.value": 52.15078686541882, "per_token.latency.mean": 0.019173916261027178, "per_token.throughput.value": 52.154186259413045 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "0a7af19f4dc868bafc82f35eb7e8d13bac87a594": { "metrics": { "prefill.latency.mean": 0.02142367935180664, "prefill.throughput.value": 326.7412606887106, "decode.latency.mean": 2.5716962890625, "decode.throughput.value": 49.3837474277716, "per_token.latency.mean": 0.020248555844224345, "per_token.throughput.value": 49.38623809486334 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "0a7af19f4dc868bafc82f35eb7e8d13bac87a594": { "metrics": { "prefill.latency.mean": 0.014151840209960936, "prefill.throughput.value": 494.6353192338173, "decode.latency.mean": 1.564446533203125, "decode.throughput.value": 81.17886888724406, "per_token.latency.mean": 0.012317361369846375, "per_token.throughput.value": 81.18621918879954 } } } } }