{ "chip" : "Apple M5 Pro", "elapsed_s" : 127.73304998874664, "latency_ms" : { "p50" : 5.3043329999999997, "p95" : 18.064624999999999 }, "lengths" : [ 128, 256, 512, 1024 ], "load_s" : 30.526584982872009, "questions" : 3899, "suites" : { "app.email_spam" : { "accuracy" : 0.99250000000000005, "buckets" : { "L128" : 98, "L256" : 117, "L512" : 93, "L1024" : 92 }, "dropped" : 0, "latency_ms" : { "p50" : 7.062875, "p95" : 18.560207999999999 }, "max_probability_delta_vs_reference" : 0.024192176759243011, "n" : 400, "reference_argmax_agreement" : 1, "reference_compared" : 400, "state_truncated" : 9 }, "app.guardrails_jailbreak" : { "accuracy" : 0.81000000000000005, "buckets" : { "L128" : 285, "L256" : 59, "L512" : 53, "L1024" : 3 }, "dropped" : 0, "latency_ms" : { "p50" : 4.0982079999999996, "p95" : 9.5879580000000004 }, "max_probability_delta_vs_reference" : 0.80097097158432007, "n" : 400, "reference_argmax_agreement" : 0.98999999999999999, "reference_compared" : 400, "state_truncated" : 0 }, "app.model_routing_domain" : { "accuracy" : 0.45363408521303256, "buckets" : { "L128" : 137, "L256" : 261, "L512" : 1 }, "dropped" : 0, "latency_ms" : { "p50" : 5.6524590000000003, "p95" : 6.3639999999999999 }, "max_probability_delta_vs_reference" : 0.077826619148254395, "n" : 399, "reference_argmax_agreement" : 0.97493734335839599, "reference_compared" : 399, "state_truncated" : 0 }, "app.moderation_toxicity" : { "accuracy" : 0.53500000000000003, "buckets" : { "L128" : 306, "L256" : 54, "L512" : 37, "L1024" : 3 }, "dropped" : 0, "latency_ms" : { "p50" : 4.0458749999999997, "p95" : 10.238166 }, "max_probability_delta_vs_reference" : 0.016416072845458984, "n" : 400, "reference_argmax_agreement" : 1, "reference_compared" : 400, "state_truncated" : 0 }, "app.phishing" : { "accuracy" : 0.99250000000000005, "buckets" : { "L128" : 76, "L256" : 111, "L512" : 104, "L1024" : 109 }, "dropped" : 0, "latency_ms" : { "p50" : 9.062875, "p95" : 19.319541999999998 }, "max_probability_delta_vs_reference" : 0.034952796995639801, "n" : 400, "reference_argmax_agreement" : 1, "reference_compared" : 400, "state_truncated" : 7 }, "app.rag_relevance" : { "accuracy" : 0.67500000000000004, "buckets" : { "L128" : 135, "L256" : 262, "L512" : 3 }, "dropped" : 0, "latency_ms" : { "p50" : 5.865875, "p95" : 9.0795840000000005 }, "max_probability_delta_vs_reference" : 0.030605673789978027, "n" : 400, "reference_argmax_agreement" : 0.99250000000000005, "reference_compared" : 400, "state_truncated" : 0 }, "app.support_triage" : { "accuracy" : 0.53749999999999998, "buckets" : { "L256" : 377, "L512" : 23 }, "dropped" : 0, "latency_ms" : { "p50" : 5.2802090000000002, "p95" : 8.903041 }, "max_probability_delta_vs_reference" : 0.052110105752944946, "n" : 400, "reference_argmax_agreement" : 0.99750000000000005, "reference_compared" : 400, "state_truncated" : 0 }, "jev.ag_news" : { "accuracy" : 0.93500000000000005, "buckets" : { "L128" : 368, "L256" : 32 }, "dropped" : 0, "latency_ms" : { "p50" : 4.5663749999999999, "p95" : 11.922083000000001 }, "max_probability_delta_vs_reference" : 0.055396914482116699, "n" : 400, "reference_argmax_agreement" : 1, "reference_compared" : 400, "state_truncated" : 0 }, "jev.emotion" : { "accuracy" : 0.53500000000000003, "buckets" : { "L128" : 400 }, "dropped" : 0, "latency_ms" : { "p50" : 3.8882919999999999, "p95" : 4.2419169999999999 }, "max_probability_delta_vs_reference" : 0.086301475763320923, "n" : 400, "reference_argmax_agreement" : 0.995, "reference_compared" : 400, "state_truncated" : 0 }, "massive_intent.en" : { "accuracy" : 0.65333333333333332, "buckets" : { "L256" : 300 }, "dropped" : 0, "latency_ms" : { "p50" : 5.5202499999999999, "p95" : 6.7780420000000001 }, "max_probability_delta_vs_reference" : 0.14386069774627686, "n" : 300, "reference_argmax_agreement" : 0.98666666666666669, "reference_compared" : 300, "state_truncated" : 0 } } }