laya-coreml / reports /verification-multilingual-L1024-e8.json
alexwengg's picture
Add e8 + compression + ANE profile reports
b7d5f10 verified
Raw History Blame Contribute Delete
18.3 kB
{
"package": "laya_multilingual_e8_L1024_options32.mlpackage",
"package_files_sha256": {
"Data/com.apple.CoreML/model.mlmodel": "1be66f2b54a08ea6a07f93c195777c4b105f7b2d9c09e47983ffecbb9f986f49",
"Data/com.apple.CoreML/weights/weight.bin": "441cefaa5768572327ba89214566c6cb9a27aaacd479523b453f442c62f08eb2",
"Manifest.json": "3dba274c81ab7dc9e42a34cbf43ad6af5329692dc2d3c3e2ce21a78ee73cbea9"
},
"source_repo": "convaiinnovations/laya",
"source_revision": "1c5edc17a7acd8701df6fc341c0d179f1c62c982",
"variant": "multilingual",
"length": 1024,
"max_options": 32,
"precision": "e8",
"package_bytes": 452793209,
"environment": {
"chip": "Apple M5 Pro",
"os": "27.0",
"python": "3.12.13",
"torch": "2.7.0",
"coremltools": "9.0"
},
"gates": {
"argmax_agreement": "100%",
"max_probability_error": 0.02,
"max_action_probability_error": 0.02,
"finite": true,
"padding_masked": true
},
"cases_sha256": "4de4e34309c421913524209b62bb7c8a0a232928bd228ec12e884347ce65eeed",
"runs": {
"ALL": {
"compute_units": "ALL",
"load_seconds": 6.339008749986533,
"questions": 16,
"argmax_agreements": 16,
"max_probability_error": 0.005797207355499268,
"max_action_probability_error": 0.0,
"latency_ms": {
"p50": 17.89439597632736,
"p95": 18.36130567244254,
"min": 17.64220802579075,
"max": 23.747125000227243
},
"skipped": [],
"rows": [
{
"case": "tetris_clean",
"question": "clean",
"type": "noul",
"tokens": 80,
"options": 2,
"argmax_agrees": true,
"coreml_argmax": 0,
"reference_argmax": 0,
"max_probability_error": 0.0025831088423728943,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.00006103515625,
"median_ms": 17.95933351968415
},
{
"case": "tetris_flat",
"question": "clean",
"type": "noul",
"tokens": 84,
"options": 2,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.0004474017769098282,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.9998931884765625,
"median_ms": 17.946458974620327
},
{
"case": "tetris_lane",
"question": "lane",
"type": "choice",
"tokens": 62,
"options": 4,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.0013618916273117065,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.9998779296875,
"median_ms": 17.91014600894414
},
{
"case": "tetris_lane",
"question": "risk",
"type": "score",
"tokens": 64,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.0008780807256698608,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.9998779296875,
"median_ms": 17.874291486805305
},
{
"case": "form_field",
"question": "action",
"type": "choice",
"tokens": 104,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 0,
"reference_argmax": 0,
"max_probability_error": 0.005797207355499268,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.999755859375,
"median_ms": 17.784750496502966
},
{
"case": "support_triage",
"question": "intent",
"type": "choice",
"tokens": 62,
"options": 4,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 5.608797073364258e-05,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.0000473260879517,
"median_ms": 18.000207986915484
},
{
"case": "support_triage",
"question": "frustration",
"type": "score",
"tokens": 69,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.002427428960800171,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.99993896484375,
"median_ms": 17.78575050411746
},
{
"case": "support_triage",
"question": "churn",
"type": "noul",
"tokens": 71,
"options": 2,
"argmax_agrees": true,
"coreml_argmax": 0,
"reference_argmax": 0,
"max_probability_error": 0.00012970063835382462,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.000091552734375,
"median_ms": 17.90124998660758
},
{
"case": "zh",
"question": "intent",
"type": "choice",
"tokens": 37,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 0,
"reference_argmax": 0,
"max_probability_error": 0.00025540590286254883,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.00022292137146,
"median_ms": 17.889228998683393
},
{
"case": "zh",
"question": "urgent",
"type": "noul",
"tokens": 48,
"options": 2,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.0006077289581298828,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.999908447265625,
"median_ms": 17.80060399323702
},
{
"case": "ja",
"question": "intent",
"type": "choice",
"tokens": 45,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 0,
"reference_argmax": 0,
"max_probability_error": 0.00017899274826049805,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.0001795291900635,
"median_ms": 17.893854004796594
},
{
"case": "de",
"question": "intent",
"type": "choice",
"tokens": 44,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 0,
"reference_argmax": 0,
"max_probability_error": 0.0004958007484674454,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.0000476837158203,
"median_ms": 17.79702099156566
},
{
"case": "many_options",
"question": "lane",
"type": "choice",
"tokens": 157,
"options": 20,
"argmax_agrees": true,
"coreml_argmax": 3,
"reference_argmax": 3,
"max_probability_error": 0.0021624043583869934,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.99993896484375,
"median_ms": 17.815499508287758
},
{
"case": "guard",
"question": "injection",
"type": "noul",
"tokens": 48,
"options": 2,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.004502624273300171,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.999755859375,
"median_ms": 18.24512452003546
},
{
"case": "long_state",
"question": "ready",
"type": "noul",
"tokens": 174,
"options": 2,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.0003440380096435547,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.9998931884765625,
"median_ms": 17.888771020807326
},
{
"case": "long_state",
"question": "owner",
"type": "choice",
"tokens": 158,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 2,
"reference_argmax": 2,
"max_probability_error": 0.0009380578994750977,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.999786376953125,
"median_ms": 17.976166476728395
}
],
"passed": true
},
"CPU_AND_NE": {
"compute_units": "CPU_AND_NE",
"load_seconds": 11.53603254200425,
"questions": 16,
"argmax_agreements": 16,
"max_probability_error": 0.013535827398300171,
"max_action_probability_error": 0.0,
"latency_ms": {
"p50": 80.1227499905508,
"p95": 81.49640440824443,
"min": 79.91241605486721,
"max": 92.36383298411965
},
"skipped": [],
"rows": [
{
"case": "tetris_clean",
"question": "clean",
"type": "noul",
"tokens": 80,
"options": 2,
"argmax_agrees": true,
"coreml_argmax": 0,
"reference_argmax": 0,
"max_probability_error": 0.0022779330611228943,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.000244140625,
"median_ms": 80.32110400381498
},
{
"case": "tetris_flat",
"question": "clean",
"type": "noul",
"tokens": 84,
"options": 2,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.0005542133003473282,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.999786376953125,
"median_ms": 80.10449950234033
},
{
"case": "tetris_lane",
"question": "lane",
"type": "choice",
"tokens": 62,
"options": 4,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.0025046318769454956,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.9996337890625,
"median_ms": 80.17289600684308
},
{
"case": "tetris_lane",
"question": "risk",
"type": "score",
"tokens": 64,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.0018546432256698608,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.0001220703125,
"median_ms": 80.06041651242413
},
{
"case": "form_field",
"question": "action",
"type": "choice",
"tokens": 104,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 0,
"reference_argmax": 0,
"max_probability_error": 0.006285488605499268,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.99951171875,
"median_ms": 80.15079150209203
},
{
"case": "support_triage",
"question": "intent",
"type": "choice",
"tokens": 62,
"options": 4,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.0004321932792663574,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.9995424151420593,
"median_ms": 80.05002100253478
},
{
"case": "support_triage",
"question": "frustration",
"type": "score",
"tokens": 69,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.0009625852108001709,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.9998779296875,
"median_ms": 80.20041699637659
},
{
"case": "support_triage",
"question": "churn",
"type": "noul",
"tokens": 71,
"options": 2,
"argmax_agrees": true,
"coreml_argmax": 0,
"reference_argmax": 0,
"max_probability_error": 9.155366569757462e-05,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.0000534057617188,
"median_ms": 80.10068751173094
},
{
"case": "zh",
"question": "intent",
"type": "choice",
"tokens": 37,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 0,
"reference_argmax": 0,
"max_probability_error": 0.00023287534713745117,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.9997386932373047,
"median_ms": 80.0608120043762
},
{
"case": "zh",
"question": "urgent",
"type": "noul",
"tokens": 48,
"options": 2,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.0005619749426841736,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.0004425048828125,
"median_ms": 81.02227098424919
},
{
"case": "ja",
"question": "intent",
"type": "choice",
"tokens": 45,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 0,
"reference_argmax": 0,
"max_probability_error": 0.00017899274826049805,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.0001797676086426,
"median_ms": 80.29522901051678
},
{
"case": "de",
"question": "intent",
"type": "choice",
"tokens": 44,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 0,
"reference_argmax": 0,
"max_probability_error": 0.0009060502052307129,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.0005512237548828,
"median_ms": 80.08233297732659
},
{
"case": "many_options",
"question": "lane",
"type": "choice",
"tokens": 157,
"options": 20,
"argmax_agrees": true,
"coreml_argmax": 3,
"reference_argmax": 3,
"max_probability_error": 0.0027727559208869934,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 1.00018310546875,
"median_ms": 80.27691699680872
},
{
"case": "guard",
"question": "injection",
"type": "noul",
"tokens": 48,
"options": 2,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.013535827398300171,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.99951171875,
"median_ms": 80.1208334742114
},
{
"case": "long_state",
"question": "ready",
"type": "noul",
"tokens": 174,
"options": 2,
"argmax_agrees": true,
"coreml_argmax": 1,
"reference_argmax": 1,
"max_probability_error": 0.0003440380096435547,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.999847412109375,
"median_ms": 80.04837500629947
},
{
"case": "long_state",
"question": "owner",
"type": "choice",
"tokens": 158,
"options": 3,
"argmax_agrees": true,
"coreml_argmax": 2,
"reference_argmax": 2,
"max_probability_error": 0.0009380578994750977,
"max_action_probability_error": 0.0,
"finite": true,
"padding_masked": true,
"probability_sum": 0.9996337890625,
"median_ms": 80.1351044792682
}
],
"passed": true
}
},
"passed": true
}