| { |
| "status": "complete_verified_paired_evaluation", |
| "qualification_sha256": "cf2d73c477bf49cb5ff9c3b20a1992e49fb2db57f22daffca5616bd5afddd8c5", |
| "production_ready": false, |
| "totals": { |
| "mlx": { |
| "en": { |
| "errors": 165, |
| "reference_units": 800, |
| "substitutions": 62, |
| "deletions": 82, |
| "insertions": 21, |
| "hits": 656, |
| "error_rate": 0.20625 |
| }, |
| "zh": { |
| "errors": 4, |
| "reference_units": 117, |
| "substitutions": 3, |
| "deletions": 0, |
| "insertions": 1, |
| "hits": 114, |
| "error_rate": 0.03418803418803419 |
| } |
| }, |
| "original": { |
| "en": { |
| "errors": 149, |
| "reference_units": 800, |
| "substitutions": 52, |
| "deletions": 76, |
| "insertions": 21, |
| "hits": 672, |
| "error_rate": 0.18625 |
| }, |
| "zh": { |
| "errors": 3, |
| "reference_units": 117, |
| "substitutions": 2, |
| "deletions": 0, |
| "insertions": 1, |
| "hits": 115, |
| "error_rate": 0.02564102564102564 |
| } |
| } |
| }, |
| "aligned_word_timing": { |
| "mlx": { |
| "matched_units": 327, |
| "ambiguous_matched_units": 1, |
| "p50_seconds": 0.5714282090272889, |
| "p95_seconds": 0.7244641249938866, |
| "maximum_seconds": 0.8183255420192381, |
| "matched_units_above_one_second": 0, |
| "unambiguous_units_above_one_second": 0, |
| "all_reference_units_within_one_second_proven": false |
| } |
| }, |
| "cases": [ |
| { |
| "id": "ami-ES2004a", |
| "language": "en", |
| "arms": { |
| "mlx": { |
| "text": "You know, I like to do the lights, so people can see that. But better,\nAm I supposed to be standing up there? So you've got both of these clipped on.\nI'm going to answer me. Yeah, I've got. Both of them. Okay.\nEverybody. Bye. I'm Sarah. Project manager. And this is our first meeting. Surprisingly enough. Okay, this is our agenda. And we will. Do some stuff. Get to know each other a bit better. Feel more comfortable with each other. Then we'll go. Do tool training. Talk about the project plan. Discuss our own ideas and everything. And we've got 25 minutes to do that. As far as I can understand. Now, we're developing a remote control. Which you probably already know. We want it to be original. Something that's people haven't thought of. It's not out in the shops. Trendy. Appealing to a wide market. But. You know. Not a hunk of metal. And user friendly. Granny's to kids. Maybe even pooches should be able to use it. Okay. First is the functional design. This is where we all go off and do our individual work. What needs need to be fulfilled by the product. What effect the product has to have? And how it's actually going to do that? Conceptual design. What we're thinking. How it's going to go. And then the detailed design. How we're actually going to put it into practice and make it work.\nOkay. Right. We're going to practice with a pen. And draw our favorite animal on the whiteboard. I'll go first. And sum up the characteristics of that animal. So.\nOkay. At least space for everyone else.\nWe're running out of blue.\nOkay. I'm not going to ask you to guess. I'm going to tell you that it's supposed to be a tiger.\nAnd I see them as majestic,\nAnd. Independent.\nAnd proud.\nNo. Who would like to go next? Me. Okay. Got.\nWith this confront, there you go, power. Oh, yep. Thank you.\nMaybe you can get so I'm trying to make.\nA kind of dog.", |
| "errors": 115, |
| "reference_units": 432, |
| "error_rate": 0.2662037037037037, |
| "memory": { |
| "os_peak_process_rss_bytes": 2338947072, |
| "os_peak_memory_footprint_bytes": 3111176664, |
| "footprint_available": true, |
| "rss_source_unit": "bytes", |
| "elapsed_seconds": 317.58, |
| "user_seconds": 56.33, |
| "system_seconds": 43.3, |
| "scope": "Fresh native child process including loading and inference; excludes Python harness. RSS and footprint are separate measurements and must not be summed." |
| }, |
| "word_timing": { |
| "measured": true, |
| "actual_arrival_paced": true, |
| "timing_basis": "Independent host receipt of flushed JSONL since paced20ms producer start; all source words retained", |
| "correctly_matched_units": 327, |
| "ambiguous_matched_units": 1, |
| "ambiguity_method": "all minimum unit-cost Levenshtein paths, without timing-based selection", |
| "unambiguous_matched_units": 326, |
| "unambiguous_maximum_seconds": 0.8183255420192381, |
| "unmatched_reference_units": 105, |
| "matched_reference_fraction": 0.7569444444444444, |
| "min_seconds": -1.6167134159724128, |
| "p50_seconds": 0.5714282090272889, |
| "p95_seconds": 0.7244641249938866, |
| "maximum_seconds": 0.8183255420192381, |
| "matched_units_above_one_second": 0, |
| "all_matched_units_within_one_second": null, |
| "whole_release_pass": false, |
| "limitation": "Lag summaries use one deterministic text alignment and are descriptive when alternative optimal alignments exist. Only unambiguous matches can support per-word timing claims. Omissions/substitutions remain errors, never zero-latency successes. Published reference timing uncertainty and overlapping-speaker word order are not independently corrected." |
| }, |
| "longest_deleted_reference_run": 18, |
| "max_queue_seconds": 0.7145231250217421, |
| "mature_windows": 718, |
| "mature_work_rtf": 0.8615857967648826 |
| }, |
| "original": { |
| "text": "You know, I like to dim the light, so people can see that. Have a better. I'm Am I supposed to be standing up there? We've got both of these clipped on. I'm going to answer me. Yeah, I've got. Both of them. Okay. Everybody, hi. I'm Sarah. Project manager. And this is our first meeting. Surprisingly enough. Okay, this is our agenda. Um, we will. Do some stuff. Get to know each other a bit better. To feel more comfortable with each other. Then. We'll go. Do tool training. Talk about the project plan. Discuss our own ideas and everything. And we've got 25 minutes to do that. As far as I can understand. Now, we're developing a remote control, which you probably already know. We want it to be original. Something that's people haven't thought of. It's not out in the shops. Trendy. Appealing to a wide market. But you know, not a hunk of metal. And user-friendly. Granny's to kids. Maybe even pooches should be able to use it. Okay. First is the functional design. This is where we all go off and do our individual work. What needs need to be fulfilled by the product. What effect the product has to have? And how it's actually going to do that? Conceptual design. What we're thinking. How it's going to go? And then the detailed design. How we're actually going to put it into practice and make it work. Okay. Right. We're going to practice with the pens. And draw our favorite animal on the whiteboard. I'll go first. And sum up the characteristics of that animal. So, Okay. At least space for everyone else. We're running out of blue. Okay. I'm not going to ask you to guess. I'm going to tell you that that's supposed to be a tiger. And I see them as majestic. And. Independent. And proud. No. Who would like to go next? Me. Okay. Got. With this confirmed. There you go, pal. Oh, yep. Thank you. Maybe you can get what I'm trying to make. A kind of dog.", |
| "errors": 109, |
| "reference_units": 432, |
| "error_rate": 0.2523148148148148 |
| } |
| } |
| }, |
| { |
| "id": "meanwhile-748OyesQy84", |
| "language": "en", |
| "arms": { |
| "mlx": { |
| "text": "Folks, if you watch the show, you know I spend most of my time right over there, carefully blending for you the day's news, you's most topical flour, eggs, milk, and butter. And straining into a fine, batter to make, delicate, yet informative comedy pancakes. Then I glaze them in the juice and zest of the most relevant midnight Valencio oranges and Dulse et al, and a fine de la main de voyage, cognac. Before, from, Baying, and basting them, tables I deserve for you the James Beard, a word-worthy crepe, Suzette, that is my nightly monologue. But sometimes, just sometimes, folks. I, wake up, and the package hold of Greyhound bus. It's being hoisted by the scrapyard claw toward the burned pit, the escape to a nearby abandoned price chopper where I scrounged for old bread scraps, busted open bags of star fruit candies, and expired eggs, chuck it all on a dirty hub cap and slap it over a tire fire before using the legs of a strained pair of sweatpants and. As oven mitts to extract and serve the demented transients pound cake of news that is my segment.", |
| "errors": 37, |
| "reference_units": 192, |
| "error_rate": 0.19270833333333334, |
| "memory": { |
| "os_peak_process_rss_bytes": 2339684352, |
| "os_peak_memory_footprint_bytes": 3103066560, |
| "footprint_available": true, |
| "rss_source_unit": "bytes", |
| "elapsed_seconds": 70.68, |
| "user_seconds": 22.78, |
| "system_seconds": 17.47, |
| "scope": "Fresh native child process including loading and inference; excludes Python harness. RSS and footprint are separate measurements and must not be summed." |
| }, |
| "word_timing": { |
| "measured": false, |
| "actual_arrival_paced": true, |
| "timing_basis": "Independent host receipt of flushed JSONL since paced20ms producer start; all source words retained", |
| "correctly_matched_units": 0, |
| "ambiguous_matched_units": 0, |
| "ambiguity_method": "all minimum unit-cost Levenshtein paths, without timing-based selection", |
| "unambiguous_matched_units": 0, |
| "unambiguous_maximum_seconds": null, |
| "unmatched_reference_units": 28, |
| "matched_reference_fraction": null, |
| "min_seconds": null, |
| "p50_seconds": null, |
| "p95_seconds": null, |
| "maximum_seconds": null, |
| "matched_units_above_one_second": null, |
| "all_matched_units_within_one_second": null, |
| "whole_release_pass": false, |
| "limitation": "Lag summaries use one deterministic text alignment and are descriptive when alternative optimal alignments exist. Only unambiguous matches can support per-word timing claims. Omissions/substitutions remain errors, never zero-latency successes. Published reference timing uncertainty and overlapping-speaker word order are not independently corrected." |
| }, |
| "longest_deleted_reference_run": 1, |
| "max_queue_seconds": 0.992105791985523, |
| "mature_windows": 387, |
| "mature_work_rtf": 0.8532934121375281 |
| }, |
| "original": { |
| "text": "Folks, if you watch this show, you know I spend most of my time right over there, carefully blending for you the day's news, the most topical flower, eggs, milk, and butter. And straining into a fine, batter to make delicate, yet informative comedy pancakes, then I glaze them in the juice and zest of the most relevant Midnight Valencia oranges and douse it all in a fine de la main de voyage, cognac. Before from being, and basting them, tables I did serve for you the James Beard Award worthy crepe sous sètte. That is my nightly monologue, but sometimes just sometimes, folks. I, wake up, and the baggage hold of Grey Hound bus is being hoisted by the scrapyard claw toward the burn pit. The escape to a nearby abandoned price chopper, where I scroung for old bread scraps, busted open bags of star fruit candies, and expired eggs, chuck it all on a dirty hub cap and slap it over a tire fire before using the legs of a strained pair of sweatpants and. As oven mitts to extract and serve the demented, transient's pound cake of news that is my segment.", |
| "errors": 29, |
| "reference_units": 192, |
| "error_rate": 0.15104166666666666 |
| } |
| } |
| }, |
| { |
| "id": "meanwhile-8prs9Pq5Xhk", |
| "language": "en", |
| "arms": { |
| "mlx": { |
| "text": "Folks, if you watch the show, and I hope you do, I spent a lot of time right over there. Tirelessly studying the lineage of the day's most important thoroughbred stories and Holsteinar headlines, working with the best trainer's money can buy to rear their comedy offspring with a hand that is stern yet gentle into the triple crown-winning equine specimen. That is my nightly monologue, but sometimes, sometimes, folks, I break into an unincorporated veterinary genetics lab. And grab whatever test tubes I can find, and then under a grow light I got from a discarded chia pet. I mix the pilfered DNA of a horse and whatever was in a tube labeled Keith Colon Extra. Slurrying the concoction with caffeine pills and a microwave of Red Bull, I scream, sing a prayer to Janus, initiator of human life and God of transformation as a halfhorse, half man, freak, seizes to life before me in the hideous collection of loose animal parts and corrupted men, tissue, that is my segment.", |
| "errors": 13, |
| "reference_units": 176, |
| "error_rate": 0.07386363636363637, |
| "memory": { |
| "os_peak_process_rss_bytes": 2342289408, |
| "os_peak_memory_footprint_bytes": 3074181544, |
| "footprint_available": true, |
| "rss_source_unit": "bytes", |
| "elapsed_seconds": 68.86, |
| "user_seconds": 22.52, |
| "system_seconds": 17.03, |
| "scope": "Fresh native child process including loading and inference; excludes Python harness. RSS and footprint are separate measurements and must not be summed." |
| }, |
| "word_timing": { |
| "measured": false, |
| "actual_arrival_paced": true, |
| "timing_basis": "Independent host receipt of flushed JSONL since paced20ms producer start; all source words retained", |
| "correctly_matched_units": 0, |
| "ambiguous_matched_units": 0, |
| "ambiguity_method": "all minimum unit-cost Levenshtein paths, without timing-based selection", |
| "unambiguous_matched_units": 0, |
| "unambiguous_maximum_seconds": null, |
| "unmatched_reference_units": 11, |
| "matched_reference_fraction": null, |
| "min_seconds": null, |
| "p50_seconds": null, |
| "p95_seconds": null, |
| "maximum_seconds": null, |
| "matched_units_above_one_second": null, |
| "all_matched_units_within_one_second": null, |
| "whole_release_pass": false, |
| "limitation": "Lag summaries use one deterministic text alignment and are descriptive when alternative optimal alignments exist. Only unambiguous matches can support per-word timing claims. Omissions/substitutions remain errors, never zero-latency successes. Published reference timing uncertainty and overlapping-speaker word order are not independently corrected." |
| }, |
| "longest_deleted_reference_run": 2, |
| "max_queue_seconds": 0.9712215409963392, |
| "mature_windows": 362, |
| "mature_work_rtf": 0.864825141212169 |
| }, |
| "original": { |
| "text": "Folks, if you watch the show, and I hope you do, I spent a lot of time right over there. Tirelessly studying the lineage of the day's most important thoroughbred stories and whole Steiner headlines, working with the best trainer's money can buy to rear their comedy offspring with a hand that is stern yet gentle into the Triple Crown-winning equine specimen. That is my nightly monologue, but sometimes, sometimes, folks, I break into an unincorporated veterinary genetics lab. And grab whatever, test tubes I can find, and then under a grow light I got from a discarded Chia pet, I mix the pilfered DNA of a horse and whatever was in a tube labeled Keith Colon Extra. Slurrying the concoction with caffeine pills and a microwave Red Bull, I scream, sing a prayer to Janice, initiator of human life and god of transformation as a half horse, half man, freak, seizes to life before me in the hideous collection of loose animal parts and corrupted men, tissue that is my segment. Meanwhile.", |
| "errors": 11, |
| "reference_units": 176, |
| "error_rate": 0.0625 |
| } |
| } |
| }, |
| { |
| "id": "BAC009S0906W0121", |
| "language": "zh", |
| "arms": { |
| "mlx": { |
| "text": "双酱会是令市场看扁人民币的情绪持续。", |
| "errors": 3, |
| "reference_units": 16, |
| "error_rate": 0.1875, |
| "memory": { |
| "os_peak_process_rss_bytes": 2338603008, |
| "os_peak_memory_footprint_bytes": 2724383072, |
| "footprint_available": true, |
| "rss_source_unit": "bytes", |
| "elapsed_seconds": 15.8, |
| "user_seconds": 8.93, |
| "system_seconds": 2.46, |
| "scope": "Fresh native child process including loading and inference; excludes Python harness. RSS and footprint are separate measurements and must not be summed." |
| }, |
| "word_timing": { |
| "measured": false, |
| "actual_arrival_paced": true, |
| "timing_basis": "Independent host receipt of flushed JSONL since paced20ms producer start; all source words retained", |
| "correctly_matched_units": 0, |
| "ambiguous_matched_units": 0, |
| "ambiguity_method": "all minimum unit-cost Levenshtein paths, without timing-based selection", |
| "unambiguous_matched_units": 0, |
| "unambiguous_maximum_seconds": null, |
| "unmatched_reference_units": 2, |
| "matched_reference_fraction": null, |
| "min_seconds": null, |
| "p50_seconds": null, |
| "p95_seconds": null, |
| "maximum_seconds": null, |
| "matched_units_above_one_second": null, |
| "all_matched_units_within_one_second": null, |
| "whole_release_pass": false, |
| "limitation": "Lag summaries use one deterministic text alignment and are descriptive when alternative optimal alignments exist. Only unambiguous matches can support per-word timing claims. Omissions/substitutions remain errors, never zero-latency successes. Published reference timing uncertainty and overlapping-speaker word order are not independently corrected." |
| }, |
| "longest_deleted_reference_run": 0, |
| "max_queue_seconds": 0.7700854999474247, |
| "mature_windows": 0, |
| "mature_work_rtf": null |
| }, |
| "original": { |
| "text": "双降会是令市场看扁人民币的情绪持续。", |
| "errors": 2, |
| "reference_units": 16, |
| "error_rate": 0.125 |
| } |
| } |
| }, |
| { |
| "id": "BAC009S0906W0122", |
| "language": "zh", |
| "arms": { |
| "mlx": { |
| "text": "人民币未来贬值压力依然较大。", |
| "errors": 0, |
| "reference_units": 13, |
| "error_rate": 0.0, |
| "memory": { |
| "os_peak_process_rss_bytes": 2338701312, |
| "os_peak_memory_footprint_bytes": 2606221664, |
| "footprint_available": true, |
| "rss_source_unit": "bytes", |
| "elapsed_seconds": 13.89, |
| "user_seconds": 8.39, |
| "system_seconds": 2.03, |
| "scope": "Fresh native child process including loading and inference; excludes Python harness. RSS and footprint are separate measurements and must not be summed." |
| }, |
| "word_timing": { |
| "measured": false, |
| "actual_arrival_paced": true, |
| "timing_basis": "Independent host receipt of flushed JSONL since paced20ms producer start; all source words retained", |
| "correctly_matched_units": 0, |
| "ambiguous_matched_units": 0, |
| "ambiguity_method": "all minimum unit-cost Levenshtein paths, without timing-based selection", |
| "unambiguous_matched_units": 0, |
| "unambiguous_maximum_seconds": null, |
| "unmatched_reference_units": 0, |
| "matched_reference_fraction": null, |
| "min_seconds": null, |
| "p50_seconds": null, |
| "p95_seconds": null, |
| "maximum_seconds": null, |
| "matched_units_above_one_second": null, |
| "all_matched_units_within_one_second": null, |
| "whole_release_pass": false, |
| "limitation": "Lag summaries use one deterministic text alignment and are descriptive when alternative optimal alignments exist. Only unambiguous matches can support per-word timing claims. Omissions/substitutions remain errors, never zero-latency successes. Published reference timing uncertainty and overlapping-speaker word order are not independently corrected." |
| }, |
| "longest_deleted_reference_run": 0, |
| "max_queue_seconds": 0.9315534170009192, |
| "mature_windows": 0, |
| "mature_work_rtf": null |
| }, |
| "original": { |
| "text": "人民币未来贬值压力依然较大。", |
| "errors": 0, |
| "reference_units": 13, |
| "error_rate": 0.0 |
| } |
| } |
| }, |
| { |
| "id": "BAC009S0907W0121", |
| "language": "zh", |
| "arms": { |
| "mlx": { |
| "text": "也不代表开发商资金面已经不再紧张。", |
| "errors": 0, |
| "reference_units": 16, |
| "error_rate": 0.0, |
| "memory": { |
| "os_peak_process_rss_bytes": 2339979264, |
| "os_peak_memory_footprint_bytes": 2628127072, |
| "footprint_available": true, |
| "rss_source_unit": "bytes", |
| "elapsed_seconds": 15.24, |
| "user_seconds": 8.77, |
| "system_seconds": 2.32, |
| "scope": "Fresh native child process including loading and inference; excludes Python harness. RSS and footprint are separate measurements and must not be summed." |
| }, |
| "word_timing": { |
| "measured": false, |
| "actual_arrival_paced": true, |
| "timing_basis": "Independent host receipt of flushed JSONL since paced20ms producer start; all source words retained", |
| "correctly_matched_units": 0, |
| "ambiguous_matched_units": 0, |
| "ambiguity_method": "all minimum unit-cost Levenshtein paths, without timing-based selection", |
| "unambiguous_matched_units": 0, |
| "unambiguous_maximum_seconds": null, |
| "unmatched_reference_units": 0, |
| "matched_reference_fraction": null, |
| "min_seconds": null, |
| "p50_seconds": null, |
| "p95_seconds": null, |
| "maximum_seconds": null, |
| "matched_units_above_one_second": null, |
| "all_matched_units_within_one_second": null, |
| "whole_release_pass": false, |
| "limitation": "Lag summaries use one deterministic text alignment and are descriptive when alternative optimal alignments exist. Only unambiguous matches can support per-word timing claims. Omissions/substitutions remain errors, never zero-latency successes. Published reference timing uncertainty and overlapping-speaker word order are not independently corrected." |
| }, |
| "longest_deleted_reference_run": 0, |
| "max_queue_seconds": 0.7304101249976087, |
| "mature_windows": 0, |
| "mature_work_rtf": null |
| }, |
| "original": { |
| "text": "也不代表开发商资金面已经不再紧张。", |
| "errors": 0, |
| "reference_units": 16, |
| "error_rate": 0.0 |
| } |
| } |
| }, |
| { |
| "id": "BAC009S0907W0122", |
| "language": "zh", |
| "arms": { |
| "mlx": { |
| "text": "背后可能蕴含着开发商更多的窘境。", |
| "errors": 0, |
| "reference_units": 15, |
| "error_rate": 0.0, |
| "memory": { |
| "os_peak_process_rss_bytes": 2339602432, |
| "os_peak_memory_footprint_bytes": 2610104672, |
| "footprint_available": true, |
| "rss_source_unit": "bytes", |
| "elapsed_seconds": 14.05, |
| "user_seconds": 8.49, |
| "system_seconds": 2.1, |
| "scope": "Fresh native child process including loading and inference; excludes Python harness. RSS and footprint are separate measurements and must not be summed." |
| }, |
| "word_timing": { |
| "measured": false, |
| "actual_arrival_paced": true, |
| "timing_basis": "Independent host receipt of flushed JSONL since paced20ms producer start; all source words retained", |
| "correctly_matched_units": 0, |
| "ambiguous_matched_units": 0, |
| "ambiguity_method": "all minimum unit-cost Levenshtein paths, without timing-based selection", |
| "unambiguous_matched_units": 0, |
| "unambiguous_maximum_seconds": null, |
| "unmatched_reference_units": 0, |
| "matched_reference_fraction": null, |
| "min_seconds": null, |
| "p50_seconds": null, |
| "p95_seconds": null, |
| "maximum_seconds": null, |
| "matched_units_above_one_second": null, |
| "all_matched_units_within_one_second": null, |
| "whole_release_pass": false, |
| "limitation": "Lag summaries use one deterministic text alignment and are descriptive when alternative optimal alignments exist. Only unambiguous matches can support per-word timing claims. Omissions/substitutions remain errors, never zero-latency successes. Published reference timing uncertainty and overlapping-speaker word order are not independently corrected." |
| }, |
| "longest_deleted_reference_run": 0, |
| "max_queue_seconds": 0.7486660419949329, |
| "mature_windows": 0, |
| "mature_work_rtf": null |
| }, |
| "original": { |
| "text": "背后可能蕴含着开发商更多的窘境。", |
| "errors": 0, |
| "reference_units": 15, |
| "error_rate": 0.0 |
| } |
| } |
| }, |
| { |
| "id": "BAC009S0908W0121", |
| "language": "zh", |
| "arms": { |
| "mlx": { |
| "text": "将进一步提振港，需购房者入市信心。", |
| "errors": 1, |
| "reference_units": 15, |
| "error_rate": 0.06666666666666667, |
| "memory": { |
| "os_peak_process_rss_bytes": 2339831808, |
| "os_peak_memory_footprint_bytes": 2688649544, |
| "footprint_available": true, |
| "rss_source_unit": "bytes", |
| "elapsed_seconds": 17.25, |
| "user_seconds": 9.25, |
| "system_seconds": 2.81, |
| "scope": "Fresh native child process including loading and inference; excludes Python harness. RSS and footprint are separate measurements and must not be summed." |
| }, |
| "word_timing": { |
| "measured": false, |
| "actual_arrival_paced": true, |
| "timing_basis": "Independent host receipt of flushed JSONL since paced20ms producer start; all source words retained", |
| "correctly_matched_units": 0, |
| "ambiguous_matched_units": 0, |
| "ambiguity_method": "all minimum unit-cost Levenshtein paths, without timing-based selection", |
| "unambiguous_matched_units": 0, |
| "unambiguous_maximum_seconds": null, |
| "unmatched_reference_units": 1, |
| "matched_reference_fraction": null, |
| "min_seconds": null, |
| "p50_seconds": null, |
| "p95_seconds": null, |
| "maximum_seconds": null, |
| "matched_units_above_one_second": null, |
| "all_matched_units_within_one_second": null, |
| "whole_release_pass": false, |
| "limitation": "Lag summaries use one deterministic text alignment and are descriptive when alternative optimal alignments exist. Only unambiguous matches can support per-word timing claims. Omissions/substitutions remain errors, never zero-latency successes. Published reference timing uncertainty and overlapping-speaker word order are not independently corrected." |
| }, |
| "longest_deleted_reference_run": 0, |
| "max_queue_seconds": 0.808853250013664, |
| "mature_windows": 0, |
| "mature_work_rtf": null |
| }, |
| "original": { |
| "text": "将进一步提振岗，需购房者入市信心。", |
| "errors": 1, |
| "reference_units": 15, |
| "error_rate": 0.06666666666666667 |
| } |
| } |
| }, |
| { |
| "id": "BAC009S0908W0122", |
| "language": "zh", |
| "arms": { |
| "mlx": { |
| "text": "加速今年楼市成交复苏回暖。", |
| "errors": 0, |
| "reference_units": 12, |
| "error_rate": 0.0, |
| "memory": { |
| "os_peak_process_rss_bytes": 2338963456, |
| "os_peak_memory_footprint_bytes": 2624457056, |
| "footprint_available": true, |
| "rss_source_unit": "bytes", |
| "elapsed_seconds": 14.37, |
| "user_seconds": 8.63, |
| "system_seconds": 2.2, |
| "scope": "Fresh native child process including loading and inference; excludes Python harness. RSS and footprint are separate measurements and must not be summed." |
| }, |
| "word_timing": { |
| "measured": false, |
| "actual_arrival_paced": true, |
| "timing_basis": "Independent host receipt of flushed JSONL since paced20ms producer start; all source words retained", |
| "correctly_matched_units": 0, |
| "ambiguous_matched_units": 0, |
| "ambiguity_method": "all minimum unit-cost Levenshtein paths, without timing-based selection", |
| "unambiguous_matched_units": 0, |
| "unambiguous_maximum_seconds": null, |
| "unmatched_reference_units": 0, |
| "matched_reference_fraction": null, |
| "min_seconds": null, |
| "p50_seconds": null, |
| "p95_seconds": null, |
| "maximum_seconds": null, |
| "matched_units_above_one_second": null, |
| "all_matched_units_within_one_second": null, |
| "whole_release_pass": false, |
| "limitation": "Lag summaries use one deterministic text alignment and are descriptive when alternative optimal alignments exist. Only unambiguous matches can support per-word timing claims. Omissions/substitutions remain errors, never zero-latency successes. Published reference timing uncertainty and overlapping-speaker word order are not independently corrected." |
| }, |
| "longest_deleted_reference_run": 0, |
| "max_queue_seconds": 0.8071327499947509, |
| "mature_windows": 0, |
| "mature_work_rtf": null |
| }, |
| "original": { |
| "text": "加速今年楼市成交复苏回暖。", |
| "errors": 0, |
| "reference_units": 12, |
| "error_rate": 0.0 |
| } |
| } |
| }, |
| { |
| "id": "BAC009S0912W0121", |
| "language": "zh", |
| "arms": { |
| "mlx": { |
| "text": "房地产相关领域问题频发。", |
| "errors": 0, |
| "reference_units": 11, |
| "error_rate": 0.0, |
| "memory": { |
| "os_peak_process_rss_bytes": 2338783232, |
| "os_peak_memory_footprint_bytes": 2597341536, |
| "footprint_available": true, |
| "rss_source_unit": "bytes", |
| "elapsed_seconds": 13.54, |
| "user_seconds": 8.33, |
| "system_seconds": 1.96, |
| "scope": "Fresh native child process including loading and inference; excludes Python harness. RSS and footprint are separate measurements and must not be summed." |
| }, |
| "word_timing": { |
| "measured": false, |
| "actual_arrival_paced": true, |
| "timing_basis": "Independent host receipt of flushed JSONL since paced20ms producer start; all source words retained", |
| "correctly_matched_units": 0, |
| "ambiguous_matched_units": 0, |
| "ambiguity_method": "all minimum unit-cost Levenshtein paths, without timing-based selection", |
| "unambiguous_matched_units": 0, |
| "unambiguous_maximum_seconds": null, |
| "unmatched_reference_units": 0, |
| "matched_reference_fraction": null, |
| "min_seconds": null, |
| "p50_seconds": null, |
| "p95_seconds": null, |
| "maximum_seconds": null, |
| "matched_units_above_one_second": null, |
| "all_matched_units_within_one_second": null, |
| "whole_release_pass": false, |
| "limitation": "Lag summaries use one deterministic text alignment and are descriptive when alternative optimal alignments exist. Only unambiguous matches can support per-word timing claims. Omissions/substitutions remain errors, never zero-latency successes. Published reference timing uncertainty and overlapping-speaker word order are not independently corrected." |
| }, |
| "longest_deleted_reference_run": 0, |
| "max_queue_seconds": 0.7206730420156382, |
| "mature_windows": 0, |
| "mature_work_rtf": null |
| }, |
| "original": { |
| "text": "房地产相关领域问题频发。", |
| "errors": 0, |
| "reference_units": 11, |
| "error_rate": 0.0 |
| } |
| } |
| }, |
| { |
| "id": "BAC009S0912W0122", |
| "language": "zh", |
| "arms": { |
| "mlx": { |
| "text": "东地产财经周刊新一年度审计工作报告出炉。", |
| "errors": 0, |
| "reference_units": 19, |
| "error_rate": 0.0, |
| "memory": { |
| "os_peak_process_rss_bytes": 2338914304, |
| "os_peak_memory_footprint_bytes": 2754447688, |
| "footprint_available": true, |
| "rss_source_unit": "bytes", |
| "elapsed_seconds": 15.96, |
| "user_seconds": 9.01, |
| "system_seconds": 2.55, |
| "scope": "Fresh native child process including loading and inference; excludes Python harness. RSS and footprint are separate measurements and must not be summed." |
| }, |
| "word_timing": { |
| "measured": false, |
| "actual_arrival_paced": true, |
| "timing_basis": "Independent host receipt of flushed JSONL since paced20ms producer start; all source words retained", |
| "correctly_matched_units": 0, |
| "ambiguous_matched_units": 0, |
| "ambiguity_method": "all minimum unit-cost Levenshtein paths, without timing-based selection", |
| "unambiguous_matched_units": 0, |
| "unambiguous_maximum_seconds": null, |
| "unmatched_reference_units": 0, |
| "matched_reference_fraction": null, |
| "min_seconds": null, |
| "p50_seconds": null, |
| "p95_seconds": null, |
| "maximum_seconds": null, |
| "matched_units_above_one_second": null, |
| "all_matched_units_within_one_second": null, |
| "whole_release_pass": false, |
| "limitation": "Lag summaries use one deterministic text alignment and are descriptive when alternative optimal alignments exist. Only unambiguous matches can support per-word timing claims. Omissions/substitutions remain errors, never zero-latency successes. Published reference timing uncertainty and overlapping-speaker word order are not independently corrected." |
| }, |
| "longest_deleted_reference_run": 0, |
| "max_queue_seconds": 0.7882775409962051, |
| "mature_windows": 0, |
| "mature_work_rtf": null |
| }, |
| "original": { |
| "text": "东地产财经周刊新一年度审计工作报告出炉。", |
| "errors": 0, |
| "reference_units": 19, |
| "error_rate": 0.0 |
| } |
| } |
| } |
| ], |
| "supplementary_causal_audits": [ |
| { |
| "case": "ami-ES2004a", |
| "audit_sha256": "eeca6befe004e557a9cbb4ea025818a294c32a93aed4d91ff349313e4a9de4f5", |
| "cross_reset_textual_matches": [], |
| "same_session_unambiguous": { |
| "count": 326, |
| "p95_seconds": 0.7244641249938866, |
| "max_seconds": 0.8183255420192381, |
| "above_one_second": 0 |
| } |
| } |
| ], |
| "no_post_result_selection": true, |
| "scope": "All eleven prospectively frozen cases; complete original references and identical causal boundaries. Selected batch2 MLX is paired with original CUDA BF16; different inputs from earlier batch4 qualification. Omissions and alignment uncertainty remain explicit; no automatic production acceptance." |
| } |
|
|
