Translate v2

internal proof

Model evaluation bench for the Kuku Yalanji training run. This page tracks the full v8-v20 diagnostic ladder, saved outputs, downloads, and resource profiles from each RunPod run.

Version lab Model registry

Where the model is after 48 hours

The training pipeline works. v20 trained the full candidate corpus and fixed the length-compression failure, but it also proved that scale alone is not enough. The hard problem is now product routing, mixture control, and faithful generalization, not GPU plumbing.

Saved evals only

Exact known text

lookup-first

Bible references, DB usage examples, and elder-shared sentence pairs should come from the approved source table, not generation.

Bible draft

v12.0

Best current Bible-draft fallback by heldout Bible chrF; still not faithful canonical reproduction.

Usage draft

v10.0

Best current heldout DB/general usage signal. Newer balanced runs have not beaten it.

Latest research

v20.0

Full-candidate-corpus diagnostic; length compression is fixed, but faithful generalization is still unresolved.

Latest run

Run: v20 full candidate corpus
Rows: 35,394 train / 4,433 validation / 4,397 test
Corpus shape: 20,911 source pairs expanded into tagged tasks
Training: 1h52m48s trainer time, 0.98 steps/s
RunPod cost: about $3.51

Model releaseDirection

Evaluation release

Live inference is intentionally off. Each release is reviewed through saved evaluation artifacts, exact-match counts, chrF/BLEU scores, resource usage, and downloadable model files.

Selected release: v20.0-full-candidate-corpus-gvn
Direction: eng-gvn
Saved eval rows: 15,556
Release role: Full-candidate-corpus diagnostic artifact

Current eval summary

Kuku Yalanji · v20.0-full-candidate-corpus-gvn

BLEU: 11.9
chrF: 39.4
Rows: 15556.0

This page shows reproducible eval artifacts only. Approved Bible text, database examples, and elder-shared sentence pairs should remain lookup-first in the product.

Dataset

v20.0 full candidate corpus: tagged Bible direct/reference tasks + DB usage replay + elder-shared sentence-pair replay

Training rows

35,394

GPU

NVIDIA L40S

Artifacts

model public · adapter public

Release verdict

Full-corpus v20 fixed the length-compression failure and completed the 35k-row diagnostic, but it is not a faithful general model: Bible heldout stays near 39 chrF, heldout usage is 43.26 chrF, and elder-shared sentence pairs regressed to 0/43 exact.

This was the first full-candidate-corpus run after the smaller v18/v19 replay runs.
The corpus was expanded into tagged direct/reference Bible tasks plus DB usage and elder-shared sentence-pair replay.
The result shows full-corpus scale alone is not enough; route-specific lookup and mixture control remain necessary before another expensive run.
Use v20 as a diagnostic artifact, not as the product default for faithful output.

Downloads and artifacts

Public links resolve through the model-artifact file server. Merged models are large safetensor directories.

Saved test outputs

Held-out examples from the saved evaluation artifacts for this model release.

Sources: ebible, Heldout usage predictions, Elder-shared sentence-pair predictions

BLEU: 11.9
chrF: 39.4
Rows: 15556.0

SourceModel outputReference

ebible

MAT.3.2

<translate> “Repent, for the Kingdom of Heaven is at hand!”

Nyulu yalaman, Yurra buyun-buyunmun jurrkijika, Godungku yurranin nyungu jawun-karra ngulkurrduku kujinka.

Nyulu yalaman, “Yurra jurrkijika buyun-buyunmun. Bama nyikuku Godundumun kaday. Godungku nyungu bama ngulkurrduku kujinka. Nyulu majamal wubulkuku.”

ebible

LUK.8.22#s1

<translate> Now on one of those days, he entered into a boat, himself and his disciples, and he said to them, “Let’s go over to the other side of the lake.”

Yala-yalaku Jesusangka jawun-karranji 12-bala boatbu walan. Nyulu jananda yalaman, Ngana ngubar-ngubar jalunbu dungarika.

Wawu yinyamun Jesus boatbu walan nyunguwunji jawun-karranji, yalaman, “Ngana naka-naka jalunbu dungarinka.”

ebible

1CO.10.15

<translate> I speak as to wise men. Judge what I say.

Ngayu yurranda balkalda, yurra binalmanka. Yurra binalmaka, ngayu wanyu balkal.

Yurra binal. Yurra junkaynjaku milkabu wukurril. Yurra binal ngayu junkaynjaku balkan-balkal.

ebible

1CO.10.16

<translate> The cup of blessing which we bless, isn’t it a sharing of the blood of Christ? The bread which we break, isn’t it a sharing of the body of Christ?

Ngana mayi breadmunbu nukanka, God thankim-bungal, nyulu ngananda dajin. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday. Ngana yalaku bunday.

Ngana communion nukal, ngana God thankim-bungal cupmunku. Ngana cupmun nukal, Christangka nyuluku mula dajil ngananga. Ngana mayi bread dumbarril, nukal, Christangka nyuluku bangkarr dajil ngananga. Yinyamundu ngana nyubunmal.

ebible

1CO.10.15

<bible_ref> 1CO.10.15 <eng> I speak as to wise men. Judge what I say.

Ngayu yurranda balkalda, yurra binalmanka. Yurra binalmaka, yurra nganya nyajilkuda.

Yurra binal. Yurra junkaynjaku milkabu wukurril. Yurra binal ngayu junkaynjaku balkan-balkal.

ebible

1CO.10.16

<bible_ref> 1CO.10.16 <eng> The cup of blessing which we bless, isn’t it a sharing of the blood of Christ? The bread which we break, isn’t it a sharing of the body of Christ?

Ngana God thankim-bungal, wine nukal, nyulu ngananda dajin, ngana nyungundu dajinka. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda. Ngana yalakuda.

ebible

1CO.10.12

<translate> Therefore let him who thinks he stands be careful that he doesn’t fall.

Kaki bama balu janaku bundanday, nyulu mumbarmaka. Nyulu kari daray, nyulu kari daray.

Mumbarmaka! Yurra kanbalda wadu-wadubuku milkabu wukurrinji, yalamanji, “Ngayu yala kari. Ngayu kari yalaku buyun-damal.” Mumbarmaka! Kaki yurra yalaku wadu-wadubuku milkabu wukurril, daranji, buyun-damanji.

ebible

1CO.10.20

<translate> But I say that the things which the Gentiles sacrifice, they sacrifice to demons and not to God, and I don’t desire that you would have fellowship with demons.

Ngayu balkalda. Bamangka Gentile-warrangka junjuy-junjuy minya wayjun, jana dubu-dubundu dajil-dajil. Jana Godundu kari dajil. Ngayu wawu yurra dubu-dubunji muruku bunday.

Yala ngayu wanyurrinku balkan yinya junjuy bawanka? Kaki bamangka mayi dajil yinyamba junjuymbu, jana dubu-dubundukuda dajil. Godundu kari. Dubu-dubundu. Ngayu wawu kari yurra murubuku mayi nukanka, Dubundumunji bamanji. Wawu kari yurra yalaku Dubu buyay-maninka.

ebible

1CO.10.12

<bible_ref> 1CO.10.12 <eng> Therefore let him who thinks he stands be careful that he doesn’t fall.

Kaki bama balu janaku bundanday, nyulu mumbarmaka. Nyulu kari daray, bubungu kaday.

ebible

1CO.10.20

<bible_ref> 1CO.10.20 <eng> But I say that the things which the Gentiles sacrifice, they sacrifice to demons and not to God, and I don’t desire that you would have fellowship with demons.

Ngayu balkalda. Bamangka Gentile-warrangka junjuy-junjuy minya wayjun, jana dubu-dubundu dajil-dajil. Jana Godundu kari dajil. Ngayu wawu yurra dubu-dubunji muruku bunday.

Heldout usage predictions

<glossary> junumundajin ; type=intransitive-verb ; domain=manipulation-impact <usage_example> The bananas are joined together

Banana junumundajin

Mayi banana junumundajin

Heldout usage predictions

<glossary> nyinki ; type=noun ; domain=body-part <usage_example> I saw him out of the corner of my eye

Ngayu nyungun nyinki nyajin miyilba

Ngayu nyungun nyinkibu nyajin

Elder-shared sentence-pair predictions

<translate> When Bama first came to the rainforest country of Manjal Dimbi, it was new and strange. They did not know the landscape or what was good to eat or what was poisonous.

Jana Bama kadarin bubungu Manjal Dimbinya, jana binal kari yinyaynka bubungu, jana binal kari yinyaynka bubungu, binal kari wanjarr ngulkurr nukanka, wanjarr yarbarrka buyun.

Yalayalaku Bama jakalba kadarin madjanga Manjal Dimbinga jana ngangkan ngangkarin bubuku, binal kari bubuku, binal kari wanyu mayi ngulkurr nukanka, wanyu mayi ngulkurr-kari nukanka.

Elder-shared sentence-pair predictions

<translate> An old man holding his spear and woomera came to Bama. He had long grey hair and long fingernails. His body shone like lightning.

Nyubun dingkar, nyulu yarraman, nyungu maral, nyungu minya, nyungu marra, nyungu mungka, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu mungka, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bangkarr, nyungu bang

Nyubun mukul dingkar, kalka, milbayarr maranga kujil kujil, jananda kadan. Nyungu mungka kalbay, bingaji, nyungu biri maranga kalbay, nyungu bangkarr balbajin yala balbay balbajinya.

ebible

DAN.5.13-29

<bible_ref> DAN.5.13-29 <eng> Then was Daniel brought in before the king. And the king spoke and said to Daniel, Art thou that Daniel, who art of the children of the captivity of Judah, whom the king my father brought out of Judea? I have even heard of thee, that the spirit of the gods is in thee, and that light and understanding and excellent wisdom is found in thee. And now the wise men , the astrologers, have been brought in before me, that they should read this writing, and make known to me the interpretation of it, but they could not show the interpretation of the thing: And I have heard of thee, that thou canst make interpretations, and dissolve doubts: now if thou canst read the writing, and make known to me the interpretation of it, thou shalt be clothed with scarlet, and have a chain of gold about thy neck, and shalt be the third ruler in the kingdom. Then Daniel answered and said before the king, Let thy gifts be to thyself, and give thy rewards to another; yet I will read the writing to the king, and make known to him the interpretation. O thou king, the most high God gave Nebuchadnezzar thy father a kingdom, and majesty, and glory, and honor: And for the majesty that he gave him, all people, nations, and languages, trembled and feared before him: whom he would he slew; and whom he would he kept alive; and whom he would he set up; and whom he would he put down. But when his heart was lifted up, and his mind hardened in pride, he was deposed from his kingly throne, and they took his glory from him: And he was driven from the sons of men; and his heart was made like the beasts, and his dwelling was with the wild asses: they fed him with grass like oxen, and his body was wet with the dew of heaven; till he knew that the most high God ruleth in the kingdom of men, and that he appointeth over it whomsoever he will. And thou his son, O Belshazzar, hast not humbled thy heart, though thou knewest all this; But hast lifted up thyself against the Lord of heaven; and they have brought the vessels of his house before thee, and thou, and thy lords, thy wives, and thy concubines, have drank wine in them; and thou hast praised the gods of silver, and gold, of brass, iron, wood, and stone, which see not, nor hear, nor know: and the God in whose hand thy breath is , and whose are all thy ways, thou hast not glorified. Then was the part of the hand sent from him; and this writing was written. And this is the writing that was written, MENE, MENE, TEKEL, UPHARSIN. This is the interpretation of the thing: MENE; God hath numbered thy kingdom, and finished it. TEKEL; Thou art weighed in the balances, and art found wanting. PERES; Thy kingdom is divided, and given to the Medes and Persians. Then commanded Belshazzar, and they clothed Daniel with scarlet, and put a chain of gold about his neck, and made a proclamation concerning him, that he should be the third ruler in the kingdom.

Nyulu Daniel kinganda wundin. Nyulu kingangka Danielanda babajin, Daniel, yundu bama Jew-warra, yundu nganjanangka bubumun Judahmun wundinda. Ngayu yunun nyajin Godungku yunun ngulkurrduku kujil. Yundu binal bajaku, dukul ngulkurr bajaku. Yinyaynka ngayu bama wubulku kunjan, jananda balkanka yinya kuku, jananda balkanka yinya kuku manubajamanka. Yamba jana binal kari yinyaynka kukuku. Ngayu yunun nyajin binalmanka yundu yinya kuku manubajamal, kuku manubajamal. Kaki yundu yinya kuku manubajamal, and ngaykundu balkanka yinyaynka kukuku, ngayu yunun ngulkurr bajaku ngulkurrduku kujil, gold medal and necklace yalbay bajaku dajil. Kingangka Danielanda yalaman, Yundu kuku ngulkurr balka, yinya kuku manubajamal.

Jana Daniel jinbaldaku kinganda wundin partynga. Danielangka kinganda kuku milkanga-bungan, yalaman, “Yinya kuku yalaku: Godungku yunun kirin kingmanka baja. Yundu kunbayda. Nyulu bama yindu king-bungal bubumun yindumun.” Nyulu kingangka Danielanda presents dajinda, nyungun maja jirray-bungan.

ebible

ACT.13.11

<bible_ref> ACT.13.11 <eng> And now behold, the hand of the Lord is upon thee, and thou shalt be blind, not seeing the sun for a season. And immediately there fell on him a mist and a darkness; and he went about seeking some to lead him by the hand.

Yundu nguba miyil-burra. Yundu ngadingka kari nyajil wungar, juma. Yinyamun ngurrbanba, wujurrbu, wungaraburr, Stephen wubulku nyajin. Stephen dungan-dungarin, bama nubin, nyungun marabu wundin.

Nyakada. Yundu buyun bajaku, bama wubul buyun-damanka. Godungku yununin miyil-burra-bungalda, yundu warngku wubul kari nyajil baja.” Yinyamun Elymas jinbalku miyil-karimankuda. Nyulu bamakuda kunjan, nyungun marabu wundinka.

Metrics

train loss: 1.485
trainer validation bleu: 13.4
trainer validation chrf: 39.9
trainer test bleu: 13.2
trainer test chrf: 40.3
validation bible direct chrf: 39.3
validation bible direct exact: 52.0
validation bible ref chrf: 39.1
validation bible ref exact: 46.0
test bible direct chrf: 39.4
test bible direct exact: 65.0
test bible ref chrf: 39.4
test bible ref exact: 67.0
heldout usage chrf: 43.3
heldout usage exact: 3.000
heldout usage headword hit rate: 0.7024
elder sentence pair chrf: 36.2
elder sentence pair exact: 0.0000
test all chrf: 39.4
test all exact: 135.0
length ratio mean test all: 1.023

Resource profile

Average GPU: 42.9%
Max GPU: 100%
Mean VRAM: 17,287 MiB
Max VRAM: 21,914 MiB
Max power: 290.49 W
Cost class: $0.99/hr
Estimated run cost: $3.51

Version ladder

Every material training branch from the last two days is listed here so the model line is auditable. chrF is useful for comparison, but exact approved known resources stay lookup-first.

23 releases

Version	Role	Bible ref	Usage	Elder rows	Decision	Links
v24.3-joint-lexeme-dose29-s3598-20260715 v24.3-joint-lexeme-dose29-s3598-20260715	Current runtime-verified Kuku Yalanji research candidate for closed-set lexical reconstruction and separately labelled sentence drafts runtime verified research candidate	n/a	53.17	0/43	The governed training-overlapping lexical gate and every frozen development retention check passed. ADVANCE authorizes further research, not unrestricted deployment: the model remains 0/43 exact on the elder diagnostic and lacks speaker-diverse human evaluation.	eval
v21.2-claude-balanced-replay-guarded-20260714 v21.2-claude-balanced-replay-guarded-20260714	Current Kuku Yalanji research model of record: frozen v21.2 weights with validated guarded decoding; CPU inference is intentionally unloaded research only	43.84	47.96	0/43	Retain the exact published v21.2 weights and guarded decoder as reproducible research artifacts, but do not load them for public CPU inference while natural elder-register transfer remains unsolved. The homepage uses separately labelled dictionary-context prompting until a stronger RunPod candidate clears the frozen linguistic and safety gates and receives appropriate speaker review.	model eval
v23.0-attested-narrative-adaptation-failed v23.0-attested-narrative-adaptation-failed	Rejected three-seed attested-narrative adaptation retained as compact negative evidence negative result	n/a	n/a	n/a	Do not promote. Seed 73 improved the held-out Text 3 corpus chrF++ point estimate by 0.6414, below the preregistered +1.0 floor; its paired interval crossed zero, untagged synthetic repetition worsened, and isolated lexical reconstruction fell to 46/297.	eval
v22.0-step-matched-replay-3120-failed v22.0-step-matched-replay-3120-failed	Rejected one-variable replay-exposure experiment retained as negative evidence negative result	n/a	n/a	n/a	Do not promote. Step 3,120 failed the frozen greedy checkpoint gate because tagged synthetic repeated-segment rows were 37 against a maximum of 25, and paired comparison showed lower tagged synthetic, untagged synthetic, and dictionary-usage agreement than the final v21.2 step 4,155 checkpoint.	eval
v21.2-claude-balanced-replay-v2-candidate v21.2-claude-balanced-replay-v2-candidate	Historical v21.2 weights release; superseded for serving by the guarded-policy release without changing weights research only	43.91	47.14	0/43	Balanced replay preserves synthetic performance while materially recovering dictionary-usage and Bible retention relative to v21.1. Natural elder-register transfer remains unsolved, and synthetic morpheme loops remain a hard safety failure; serve only as a clearly labelled research draft pending speaker review.	model eval
v21.1-codex-synthetic-direct-v2-candidate v21.1-codex-synthetic-direct-v2-candidate	Kuku Yalanji translation model v2 candidate research only	28.76	34.40	0/43	The model learned the synthetic treatment and remained stable without the task tag, but elder-shared, dictionary-usage, and Bible controls regressed severely. Preserve as a research artifact; keep retrieval-first routing and require balanced replay plus speaker review before promotion.
v20.0-full-candidate-corpus-gvn v20.0-full-candidate-corpus-gvn	Full-candidate-corpus diagnostic artifact internal proof	39.39	43.26	0/43	Full-corpus v20 fixed the length-compression failure and completed the 35k-row diagnostic, but it is not a faithful general model: Bible heldout stays near 39 chrF, heldout usage is 43.26 chrF, and elder-shared sentence pairs regressed to 0/43 exact.	model eval
v19.0-balanced-replay-from-v12-gvn v19.0-balanced-replay-from-v12-gvn	Balanced replay artifact internal proof	44.29	53.03	29/43	Restores near-v12 Bible heldout while adding elder sentence-pair/usage signal, but does not replace v10 usage or v12 Bible routing.	model eval
v18.0-usage-elder-sentence-continuation-from-v10 v18.0-usage-elder-sentence-continuation-from-v10	Elder sentence-pair memorization proof artifact internal proof	41.25	54.52	43/43	Exactly memorized all elder-shared sentence pairs, but regressed Bible and did not beat v10 on heldout usage.	model eval
v15.0-soft-lexical-hint-bible-gvn-token v15.0-soft-lexical-hint-bible-gvn-token	Lexical-hint diagnostic negative result	43.72	n/a	n/a	Reproduced train rows strongly, but did not beat v12 on heldout Bible.	model eval
v13.0-retrieval-context-bible-gvn-token v13.0-retrieval-context-bible-gvn-token	Retrieval-prefix diagnostic negative result	29.71	n/a	n/a	Retrieval context as an NLLB source prefix was harmful; heldout Bible dropped sharply.	model eval
v12.0-tagged-direct-plus-reference-bible-gvn-token v12.0-tagged-direct-plus-reference-bible-gvn-token	Current Bible draft fallback route candidate	44.36	n/a	n/a	Best Bible draft heldout score before v19, but still zero exact heldout reproduction.	model eval
v11.0-byt5-bible-control-32row v11.0-byt5-bible-control-32row	Byte-model control negative result	5.71	n/a	n/a	Failed the tiny overfit/control path; NLLB/LoRA remains the proven memorizing path for now.
v10.0-tagged-bible-plus-glossary-usage-tpi v10.0-tagged-bible-plus-glossary-usage-tpi	Current usage/general draft fallback route candidate	43.29	56.18	n/a	Best heldout usage signal so far; weaker Bible than v12/v19.	model eval
v9.8-tagged-bible-plus-db-usage-tpi v9.8-tagged-bible-plus-db-usage-tpi	First DB-usage multitask diagnostic internal proof	44.04	39.47	n/a	Memorized DB train examples but did not generalize well to word-id heldout usage.	model eval
v9.7-tagged-direct-plus-reference-bible-tpi v9.7-tagged-direct-plus-reference-bible-tpi	Reference-conditioning baseline internal proof	44.34	n/a	n/a	Established tagged Bible direct/reference training as viable, but still zero exact heldout reproduction.	model eval
v8.0-diagnostic-gates-summary v8.0-diagnostic-gates-summary	Pipeline proof gate internal proof	n/a	n/a	n/a	Proved tokenizer/LoRA/training/merge path could memorize; shifted problem from plumbing to data/task/generalization.
0.7.0-full-tpi-proxy-1.3b 0.7.0-full-tpi-proxy-1.3b	kuku_yalanji_ebible_parallel_v0.1.0 internal proof	n/a	n/a	n/a	No verdict recorded.	model eval
0.4.0-full-gvn-token 0.4.0-full-gvn-token	kuku_yalanji_ebible_parallel_v0.1.0 internal proof	n/a	n/a	n/a	No verdict recorded.	model eval
0.1.0-mini-pilot 0.1.0-mini-pilot	kuku_yalanji_ebible_parallel_v0.1.0 internal proof	n/a	n/a	n/a	No verdict recorded.
0.1.0-smoke 0.1.0-smoke	kuku_yalanji_ebible_parallel_smoke_v0.1.0 internal proof	n/a	n/a	n/a	No verdict recorded.
0.1.0-baseline 0.1.0-baseline	kuku_yalanji_ebible_parallel_v0.1.0 training ready	n/a	n/a	n/a	No verdict recorded.
1.0.0-rc1 1.0.0-rc1	First leakage-controlled Mi'gmaq dictionary-example research candidate research only	n/a	n/a	n/a	Machine-integrity gates pass, but linguistic-quality promotion does not: frozen-test chrF++ is 21.43 with 0/742 exact matches and material lexical/content failures. Permit only a visibly warned research preview pending qualified Mi'gmaq review.	model eval