{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "48681e6e5ef08e5af8b3feb5683f044321605f1f": { "metrics": { "prefill.latency.mean": 0.019090847969055177, "prefill.throughput.value": 366.66784059809555, "decode.latency.mean": 2.3359691162109373, "decode.throughput.value": 54.36715713348153, "per_token.latency.mean": 0.018466484107518857, "per_token.throughput.value": 54.15215989018926 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "48681e6e5ef08e5af8b3feb5683f044321605f1f": { "metrics": { "prefill.latency.mean": 0.021794272422790528, "prefill.throughput.value": 321.1853033772312, "decode.latency.mean": 2.5525396728515624, "decode.throughput.value": 49.7543686982629, "per_token.latency.mean": 0.020181032497421085, "per_token.throughput.value": 49.55147860387168 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "48681e6e5ef08e5af8b3feb5683f044321605f1f": { "metrics": { "prefill.latency.mean": 0.013933776378631592, "prefill.throughput.value": 502.37637018023224, "decode.latency.mean": 1.5579072265625, "decode.throughput.value": 81.51961672340636, "per_token.latency.mean": 0.012319213772950909, "per_token.throughput.value": 81.17401146132258 } } } } }