{ "chip" : "Apple M5 Pro", "elapsed_s" : 23.682847023010254, "latency_ms" : { "p50" : 5.3144580000000001, "p95" : 18.081707999999999 }, "lengths" : [ 128, 256, 512, 1024 ], "load_s" : 2.8048709630966187, "questions" : 3899, "suites" : { "app.email_spam" : { "accuracy" : 0.99250000000000005, "buckets" : { "L128" : 98, "L256" : 117, "L512" : 93, "L1024" : 92 }, "dropped" : 0, "latency_ms" : { "p50" : 6.297625, "p95" : 18.979500000000002 }, "max_probability_delta_vs_reference" : 0.0070431232452392578, "n" : 400, "reference_argmax_agreement" : 1, "reference_compared" : 400, "state_truncated" : 9 }, "app.guardrails_jailbreak" : { "accuracy" : 0.8075, "buckets" : { "L128" : 285, "L256" : 59, "L512" : 53, "L1024" : 3 }, "dropped" : 0, "latency_ms" : { "p50" : 4.1935419999999999, "p95" : 9.8876670000000004 }, "max_probability_delta_vs_reference" : 0.01729881763458252, "n" : 400, "reference_argmax_agreement" : 0.99750000000000005, "reference_compared" : 400, "state_truncated" : 0 }, "app.model_routing_domain" : { "accuracy" : 0.44110275689223055, "buckets" : { "L128" : 137, "L256" : 261, "L512" : 1 }, "dropped" : 0, "latency_ms" : { "p50" : 5.391, "p95" : 6.2202919999999997 }, "max_probability_delta_vs_reference" : 0.014412820339202881, "n" : 399, "reference_argmax_agreement" : 1, "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" : 3.8999999999999999, "p95" : 9.4122920000000008 }, "max_probability_delta_vs_reference" : 0.0036969184875488281, "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" : 8.9892920000000007, "p95" : 18.856666000000001 }, "max_probability_delta_vs_reference" : 0.003923460841178894, "n" : 400, "reference_argmax_agreement" : 1, "reference_compared" : 400, "state_truncated" : 7 }, "app.rag_relevance" : { "accuracy" : 0.67249999999999999, "buckets" : { "L128" : 135, "L256" : 262, "L512" : 3 }, "dropped" : 0, "latency_ms" : { "p50" : 5.4204169999999996, "p95" : 6.3289999999999997 }, "max_probability_delta_vs_reference" : 0.0088013410568237305, "n" : 400, "reference_argmax_agreement" : 1, "reference_compared" : 400, "state_truncated" : 0 }, "app.support_triage" : { "accuracy" : 0.54249999999999998, "buckets" : { "L256" : 377, "L512" : 23 }, "dropped" : 0, "latency_ms" : { "p50" : 5.5369999999999999, "p95" : 9.0276250000000005 }, "max_probability_delta_vs_reference" : 0.0048961937427520752, "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" : 4.6144170000000004, "p95" : 7.2706249999999999 }, "max_probability_delta_vs_reference" : 0.010490596294403076, "n" : 400, "reference_argmax_agreement" : 1, "reference_compared" : 400, "state_truncated" : 0 }, "jev.emotion" : { "accuracy" : 0.53749999999999998, "buckets" : { "L128" : 400 }, "dropped" : 0, "latency_ms" : { "p50" : 4.0142920000000002, "p95" : 4.3915829999999998 }, "max_probability_delta_vs_reference" : 0.020582765340805054, "n" : 400, "reference_argmax_agreement" : 1, "reference_compared" : 400, "state_truncated" : 0 }, "massive_intent.en" : { "accuracy" : 0.65666666666666662, "buckets" : { "L256" : 300 }, "dropped" : 0, "latency_ms" : { "p50" : 5.3079169999999998, "p95" : 6.1710409999999998 }, "max_probability_delta_vs_reference" : 0.019814342260360718, "n" : 300, "reference_argmax_agreement" : 1, "reference_compared" : 300, "state_truncated" : 0 } } }