open-standards/aac/aac-rs/tools/ae/languages.json
Transaction Science 9e25412dc0 fix(aac first letters): the evidence as the app asks by default (slips forgiven)
The app forgives slips to a neighbouring key unless the person turns it off,
so English's evidence is measured that way: on the 252 Turk Dialogues
replies with what the partner said, 51.2% among the five sentences and
31.7% first (52.8% and 32.9% with slips not forgiven), above the 50 bar.
The catalogue's credits link where each piece is published.

Co-Authored-By: Claude Opus 5.5 <[email protected]>
2026-10-02 00:51:00 -04:00

235 lines
9.3 KiB
JSON

{
"about": "The languages first letters (abbreviation expansion) is offered in. A language is ready only when we measured it on this app's own worker (tools/ae/eval): replies from typed conversations, each given as its first letters with what the partner said before it, and the reply meant was among the five sentences offered for at least 50 in 100 of at least 200 replies. tools/ae/build_ae.py copies this into the catalogue.",
"languages": {
"am": {
"status": "not-yet",
"name": "Amharic",
"note": "Only English so far: first letters in Amharic needs its own conversations to learn from, and its own test."
},
"ar": {
"status": "not-yet",
"name": "Arabic",
"note": "Only English so far: first letters in Arabic needs its own conversations to learn from, and its own test."
},
"bn": {
"status": "not-yet",
"name": "Bengali",
"note": "Only English so far: first letters in Bengali needs its own conversations to learn from, and its own test."
},
"cs": {
"status": "not-yet",
"name": "Czech",
"note": "Only English so far: first letters in Czech needs its own conversations to learn from, and its own test."
},
"da": {
"status": "not-yet",
"name": "Danish",
"note": "Only English so far: first letters in Danish needs its own conversations to learn from, and its own test."
},
"de": {
"status": "not-yet",
"name": "German",
"note": "Only English so far: first letters in German needs its own conversations to learn from, and its own test."
},
"el": {
"status": "not-yet",
"name": "Greek",
"note": "Only English so far: first letters in Greek needs its own conversations to learn from, and its own test."
},
"en": {
"status": "ready",
"name": "English",
"model": "ae-en-qwen3-0.6b",
"measured": 51.2,
"top1": 31.7,
"items": 252,
"source": "ours (what decides): Turk Dialogues (K. Vertanen, 2017; used for testing only, never learned from), test split, the first sentence of each second turn whose first letters are at most 10, with the first turn as what the partner said (252 replies), through this app's worker on onnxruntime-web 1.30.0 WebGPU (Chrome 154, Apple M5 Max), asked as the app asks by default (punctuation where typed, slips to a neighbouring key forgiven); counted when one of the five reads as the reply, case, spacing and final punctuation aside; 2026-10-01 (tools/ae/eval/summary-2026-10-01.json)",
"note": "In our test on 252 replies from typed conversations, given as first letters with what the partner had said, the reply meant was among the five sentences for about 51 in 100, and first for about 32. Without what the partner said, about 33 in 100."
},
"es": {
"status": "not-yet",
"name": "Spanish",
"note": "Only English so far: first letters in Spanish needs its own conversations to learn from, and its own test."
},
"fa": {
"status": "not-yet",
"name": "Persian",
"note": "Only English so far: first letters in Persian needs its own conversations to learn from, and its own test."
},
"fi": {
"status": "not-yet",
"name": "Finnish",
"note": "Only English so far: first letters in Finnish needs its own conversations to learn from, and its own test."
},
"fr": {
"status": "not-yet",
"name": "French",
"note": "Only English so far: first letters in French needs its own conversations to learn from, and its own test."
},
"gu": {
"status": "not-yet",
"name": "Gujarati",
"note": "Only English so far: first letters in Gujarati needs its own conversations to learn from, and its own test."
},
"ha": {
"status": "not-yet",
"name": "Hausa",
"note": "Only English so far: first letters in Hausa needs its own conversations to learn from, and its own test."
},
"he": {
"status": "not-yet",
"name": "Hebrew",
"note": "Only English so far: first letters in Hebrew needs its own conversations to learn from, and its own test."
},
"hi": {
"status": "not-yet",
"name": "Hindi",
"note": "Only English so far: first letters in Hindi needs its own conversations to learn from, and its own test."
},
"hu": {
"status": "not-yet",
"name": "Hungarian",
"note": "Only English so far: first letters in Hungarian needs its own conversations to learn from, and its own test."
},
"id": {
"status": "not-yet",
"name": "Indonesian",
"note": "Only English so far: first letters in Indonesian needs its own conversations to learn from, and its own test."
},
"it": {
"status": "not-yet",
"name": "Italian",
"note": "Only English so far: first letters in Italian needs its own conversations to learn from, and its own test."
},
"ja": {
"status": "not-yet",
"name": "Japanese",
"note": "First letters is for languages written in letters with spaces between words; Japanese is written another way, so it needs a way of its own."
},
"kn": {
"status": "not-yet",
"name": "Kannada",
"note": "Only English so far: first letters in Kannada needs its own conversations to learn from, and its own test."
},
"ko": {
"status": "not-yet",
"name": "Korean",
"note": "Only English so far: first letters in Korean needs its own conversations to learn from, and its own test."
},
"ml": {
"status": "not-yet",
"name": "Malayalam",
"note": "Only English so far: first letters in Malayalam needs its own conversations to learn from, and its own test."
},
"mr": {
"status": "not-yet",
"name": "Marathi",
"note": "Only English so far: first letters in Marathi needs its own conversations to learn from, and its own test."
},
"ms": {
"status": "not-yet",
"name": "Malay",
"note": "Only English so far: first letters in Malay needs its own conversations to learn from, and its own test."
},
"nb": {
"status": "not-yet",
"name": "Norwegian",
"note": "Only English so far: first letters in Norwegian needs its own conversations to learn from, and its own test."
},
"nl": {
"status": "not-yet",
"name": "Dutch",
"note": "Only English so far: first letters in Dutch needs its own conversations to learn from, and its own test."
},
"pa": {
"status": "not-yet",
"name": "Punjabi",
"note": "Only English so far: first letters in Punjabi needs its own conversations to learn from, and its own test."
},
"pl": {
"status": "not-yet",
"name": "Polish",
"note": "Only English so far: first letters in Polish needs its own conversations to learn from, and its own test."
},
"pt": {
"status": "not-yet",
"name": "Portuguese",
"note": "Only English so far: first letters in Portuguese needs its own conversations to learn from, and its own test."
},
"ro": {
"status": "not-yet",
"name": "Romanian",
"note": "Only English so far: first letters in Romanian needs its own conversations to learn from, and its own test."
},
"ru": {
"status": "not-yet",
"name": "Russian",
"note": "Only English so far: first letters in Russian needs its own conversations to learn from, and its own test."
},
"sv": {
"status": "not-yet",
"name": "Swedish",
"note": "Only English so far: first letters in Swedish needs its own conversations to learn from, and its own test."
},
"sw": {
"status": "not-yet",
"name": "Swahili",
"note": "Only English so far: first letters in Swahili needs its own conversations to learn from, and its own test."
},
"ta": {
"status": "not-yet",
"name": "Tamil",
"note": "Only English so far: first letters in Tamil needs its own conversations to learn from, and its own test."
},
"te": {
"status": "not-yet",
"name": "Telugu",
"note": "Only English so far: first letters in Telugu needs its own conversations to learn from, and its own test."
},
"th": {
"status": "not-yet",
"name": "Thai",
"note": "First letters is for languages written in letters with spaces between words; Thai is written another way, so it needs a way of its own."
},
"tl": {
"status": "not-yet",
"name": "Filipino",
"note": "Only English so far: first letters in Filipino needs its own conversations to learn from, and its own test."
},
"tr": {
"status": "not-yet",
"name": "Turkish",
"note": "Only English so far: first letters in Turkish needs its own conversations to learn from, and its own test."
},
"uk": {
"status": "not-yet",
"name": "Ukrainian",
"note": "Only English so far: first letters in Ukrainian needs its own conversations to learn from, and its own test."
},
"ur": {
"status": "not-yet",
"name": "Urdu",
"note": "Only English so far: first letters in Urdu needs its own conversations to learn from, and its own test."
},
"vi": {
"status": "not-yet",
"name": "Vietnamese",
"note": "Only English so far: first letters in Vietnamese needs its own conversations to learn from, and its own test."
},
"yo": {
"status": "not-yet",
"name": "Yoruba",
"note": "Only English so far: first letters in Yoruba needs its own conversations to learn from, and its own test."
},
"zh": {
"status": "not-yet",
"name": "Chinese",
"note": "First letters is for languages written in letters with spaces between words; Chinese is written another way, so it needs a way of its own."
},
"zu": {
"status": "not-yet",
"name": "Zulu",
"note": "Only English so far: first letters in Zulu needs its own conversations to learn from, and its own test."
}
}
}