Lilly

A Bosnian translator you can type into, talk to, or point a camera at — running entirely on your own machine.

Why · Examples · Run it · Inside · Journey · Numbers · Limits · Training

---

Why I built it

I applied to a lot of internships and kept getting turned down for the same reason: I did not know Bosnian.

I could not learn a language in a week, so I built the thing I needed instead. Lilly is a translator I made for myself. In class and at work I press the microphone, say what I just heard, and get it back in English. I point the camera at a whiteboard or a sign and read it. I type what I want to say in English and Lilly gives me the Bosnian to say back.

I tried the tools that already existed. Google Translate and the rest kept getting my sentences wrong — close enough to look right, wrong enough to leave me lost in the room. They also treat Bosnian as one more entry in a South Slavic bucket, and that is not the same thing as understanding it.

So I trained the models myself. The first versions were bad — under 30% of the words on a sign, more than half the words of a sentence heard wrong — and I kept the numbers as they climbed, measured every change against the untuned model underneath it, and published all of them, including the ones that say a change did nothing. Lilly is how I show that I can work in Bosnian and that I can build something serious when I hit a wall.

The longer version is in docs/STORY.md.

---

See it work

Real output from the running app, not hand-picked from a benchmark.

Bosnian → English

You type	Lilly returns
Molim vas, možete li ponoviti? Nisam razumio zadaću.	Please, can you repeat that? I didn't understand the assignment.
Sastanak je sutra u devet u kancelariji na drugom spratu.	The meeting is tomorrow at nine in the office on the second floor.
Rok za predaju projekta je sljedeći petak.	The deadline for handing over the project is next Friday.

English → Bosnian, for saying something back

You type	Lilly returns
Could you send me the file before the meeting?	Možete li mi poslati dosje prije sastanka?
I am still learning Bosnian, please be patient with me.	Još učim bosanski, molim vas budite strpljivi sa mnom.

Speech and photographs go through the same translator, so a spoken sentence and a photographed sign come back the same way a typed one does.

---

Run it

uv venv .venv --python 3.12
uv pip install --python .venv/bin/python -r requirements.txt
.venv/bin/python scripts/fetch_models.py        # the bundle, then the reader's own weights
.venv/bin/uvicorn app.server:app --port 8000

Open http://localhost:8000. Allow the microphone, or photograph something with a đ in it.

That gives you all five abilities. The translator arrives already fine-tuned and quantised, so there is nothing to build for the forward direction, and the bundle has carried the reply direction (English in, Bosnian out, the swap button in the UI) since 8 September 2026, so fetch_models.py pulls it as translator-en-bs/. The fetch also pulls PP-OCRv6 into PaddleX's cache through the app's own reader, so the first photograph does not wait on a download, and it checks the listener it was handed against the gated one, saying so out loud if an older fetch left the closed whisper-large-v3 behind. To rebuild the reply direction yourself from the upstream base instead:

.venv/bin/python scripts/fetch_translate_base.py --direction en-bs
.venv/bin/python scripts/build_translator.py --direction en-bs

Without translator-en-bs/ the app still runs and /api/reply answers 503. python3 app/lilly.py prints which parts are installed.

Every ability runs both ways. The arrow between the two language names is a button: swap it and Lilly hears English, reads an English photograph, answers in Bosnian, and says the answer out loud. The listener and the reader are the same weights either way; the one new part is the Bosnian voice, speak-bs/, which fetch_models.py pulls from rhasspy/piper-voices rather than from the bundle. Piper has no Bosnian voice; this is the one it files under Serbian, read through Serbian phonemes (so numbers come out the Serbian way, "dve" for "dvije"), and its own card says the recordings behind it are the Sorbian Institute's Lower Sorbian data — intelligible, accented, and worth hearing before relying on. Without it the app still runs and /api/speak with "language": "bs" answers 503. How well Lilly hears and reads English has not been measured here — see What it cannot do yet.

Startup is instant because each model loads on first use. Once fetch_models.py has finished, nothing reaches the network again. python3 -m pytest tests checks the parts that need no model: the sentence splitter, the batch grouping, the correction store and the server's answers to bad input.

---

What's inside

Every ability sits behind one object and one API.

from app.lilly import lilly
lilly.translate("Dobar dan")          # Bosnian text   -> English text
lilly.reply("Good morning")           # English text   -> Bosnian text
lilly.listen("clip.m4a")              # spoken Bosnian -> Bosnian text
lilly.read("sign.jpg")                # photo          -> Bosnian text
lilly.speak("Good day", "out.wav")    # English text   -> spoken English

# and the other way round, for answering back
lilly.listen("clip.m4a", language="en")               # spoken English -> English text
lilly.speak("Dobar dan", "out.wav", language="bs")    # Bosnian text   -> spoken Bosnian
lilly.translate_audio("clip.m4a", direction="en-bs")  # spoken English -> (Bosnian, English)
lilly.translate_photo("sign.jpg", direction="en-bs")  # English photo  -> (Bosnian, English)

Endpoint	Body	Returns
POST /api/translate	{"text": "..."}	Bosnian in, English out
POST /api/reply	{"text": "..."}	English in, Bosnian out
POST /api/speech	audio upload, optional direction field	transcribes, then translates: Bosnian heard → English (bs-en, the default) or English heard → Bosnian (en-bs); the answer is {"bosnian", "english"} either way
POST /api/photo	image upload, optional direction field	reads the text off the image, then translates it, the same two ways
POST /api/speak	{"text": "...", "language": "en"}	speech as WAV; "bs" reads the Bosnian answer with the Bosnian voice
POST /api/feedback	a correction	stored for review and retraining
GET /health	—	liveness

Every request is bounded before it reaches a model — uploads by size, text by how much work it asks for, images by pixel count — because the server is written to face the open internet.

The correction button

When a translation is wrong, you press This translation is wrong, say what it should have been, and the correction is stored for review. Verified corrections go back into the training pool, so the model improves on the sentences people actually hit rather than the ones a benchmark happens to contain.

Weights it is built from

Lilly is a bundle, not a new architecture. The translator and the listener are fine-tuned here. The reader and the voice are off the shelf: the reader is PP-OCRv6, chosen over the fine-tuned EasyOCR reader by a rule written before the comparison ran, and four later attempts to fine-tune it never beat it. Full attribution and licenses are in models/lilly/NOTICE.md.

Ability	Built from	Fine-tuned here
Translate	OPUS-MT opus-mt-tc-big-zls-en (Helsinki-NLP), CTranslate2 int8	yes — LoRA merged into the weights
Reply	OPUS-MT opus-mt-tc-base-en-sh (Helsinki-NLP), CTranslate2 int8	yes — LoRA merged into the weights; cleared its gate and published 8 Sep 2026
Listen	whisper-large-v3 (OpenAI), converted to CTranslate2 int8 here	yes — LoRA. Shipped by the owner's decision, refused at its gate (see Speech); the gated faster-whisper-small fine-tune stays the baseline
Read	PaddleOCR PP-OCRv6 (PP-OCRv6_medium_det + _medium_rec, PaddlePaddle), fetched at run time; EasyOCR + CRAFT stays as the LILLY_READER=easyocr way back	no — stock, chosen by a pre-registered rule; the EasyOCR fallback is fine-tuned
Speak	English: Kokoro-82M (hexgrad). Bosnian: Piper sr_RS-serbski_institut-medium (rhasspy) — filed under Serbian, trained on the Sorbian Institute's recordings by its own card, since Piper has none for Bosnian; fetched from upstream, not bundled	no — stock weights

---

How well it works

The journey so far

The first builds were worse than anything below. No measurement of them was kept in the repository — the habit of committing a number before changing anything came later, and is now the rule — so the first column comes from the owner's own notes from that time, written as a bound. The next column is the first number that was recorded, with the file it lives in. The last is today.

	the first builds (unrecorded)	first recorded	today
Photographs — words found per photograph, the 40	< 30%	36.0% (training/RESULTS-ocr.md)	67.0%
Photographs — words found, pooled	< 10%	16.9% (63 of 373)	69.4%
Photographs — words invented that are on no sign	> 280	224	65
Speech — word error, 200 held-out clips	> 55%	38.5% (training/RESULTS-speech.md)	11.9% (whisper-large-v3, shipped by decision, refused at its gate; the gated whisper-small reads 34.9%)
Translation — BLEU on FLORES devtest, as the user sees it	< 30	37.72, with the language tag leaked into 308 of 1,012 outputs (training/RESULTS-devtest.md)	43.25, leaked into 0 (re-measured 8 Sep on a T4 after the ordinal splitter fix)
Reply, English → Bosnian — chrF2 on FLORES-200	—	58.96, the base as downloaded (training/RESULTS-en-bs.md)	60.00, fine-tuned and published 8 Sep

One recorded moment says what the early period was like: the reader scored about 75% on synthetic text and 36% the first time it was pointed at real photographs (training/RESULTS-ocr-dataset.md). The 75% was never a real number. Everything after that was measured on real photographs, real audio and held-out sentences.

Where the credit goes

The fairest thing to measure a change against is the untuned model it is built on. Every "Lilly today" figure below is measured on data the model never trained on, through the app's own code path, against exactly that base. The pass marks were written down before each run (see thresholds).

Ability	Measured on	Untuned base	Lilly today	Where it stands
Translate, Bosnian → English	1,012 FLORES devtest sentences, as the user sees them	42.08 BLEU / 67.85 chrF2 with its tag stripped; the base emits its language tag into 576 of 2,009 outputs, which the app strips since 8 Sep	43.25 BLEU / 68.10 chrF2, 0 leaks	shipped; re-measured 8 Sep on a T4 after the ordinal splitter fix
Reply, English → Bosnian	2,009 FLORES-200 pairs	29.57 BLEU / 58.96 chrF2	30.73 / 60.00; writes the Bosnian form of a contested word 99.2% of the time (base 94.3%)	cleared all four bars 8 Sep; in the bundle since 8 Sep
Listen, whisper-small	200 held-out FLEURS clips	38.5% word error	34.9% word error; Bosnian term recall 65.9% → 68.2%	the listener that cleared its gate; kept as the baseline
Listen, whisper-large-v3	the same 200, then all 925	—	11.9% word error on the 200; 14.1% on 925 against small's 39.5%	shipped since 8 Sep by the owner's decision, refused at its gate: writes Croatian forms more often (1.1% → 6.1%); closed to further looks, see below
Read	40 Commons photographs; test-v2, 132 with text	first reader: 36.0% of sign words found, 224 invented	67.0% found, 65 invented on the 40; 57.8% found, 450 invented on test-v2	shipped (PP-OCRv6, untrained)

How to read the numbers

• Word error rate (speech): the share of words heard wrong. Lower is better.
• BLEU and chrF2 (translation): how much the output overlaps a professional translation. Higher is better. chrF2 counts characters, so it is the fairer measure for a heavily inflected language, and it is the one that decides here. A difference of a point or so is noise unless a paired bootstrap says otherwise; every gain quoted above has one.
• Words found and invented (photographs): the share of the words on the signs that the reader read correctly, and how many words it produced that are on no sign at all. The second number matters as much as the first, because recall can always be bought by guessing more.
• Published: the public bundle Safak11/lilly carries exactly the builds these numbers were measured on. The publisher checks each one by content fingerprint and refuses any other; the listener goes up only under a fingerprint named on the command line, because it did not clear its gate.

Translation

Held-out FLORES-200 Bosnian–English, through the app's own path.

	the first builds (unrecorded)	first recorded	today
BLEU, 1,012 devtest pairs	< 30	37.72 — the model's language tag leaked into 308 of 1,012 outputs (training/RESULTS-devtest.md)	43.25, leaked into 0
chrF2, 1,012 devtest pairs	—	67.15	68.10

The "first recorded" column is the untuned model as downloaded. Scored with its leaked tags stripped, so the defect cannot take credit, the fine-tuning is worth +1.26 BLEU at p = 0.001 and +0.15 chrF2 at p = 0.074, which does not clear 0.05, on all 2,009 pairs (training/RESULTS-product.md; on the devtest half +1.17 / +0.25). In plain terms: it makes word-level accuracy better, it removes a defect from every third output, and it does not move chrF2 by an amount the bootstrap can see. The first builds, under 30 BLEU, did not manage any of that.

Re-measured on 8 September 2026 after the splitter fix. Until that day app.translate.Engine cut a Bosnian date into pieces (5. maja 1990. godine became three sentences) and translated each alone. The fix was pre-registered (training/PREREGISTRATION.md, "v4 — translate — ordinals") and then measured on Kaggle with both splitters on the same T4: the new rule is worth +0.89 BLEU / +0.37 chrF2 to the fine-tune and +1.05 / +0.38 to the base (p = 0.001, 167 of 2,009 rows changed), and the difference between the Mac's CPU and the T4 on the old rule is inside noise (−0.04 BLEU, p = 0.25). The figures above are the new ones; the old ones (42.49 / 67.69) are in the records, not erased.

The reply direction (English → Bosnian) was fine-tuned on 8 September and cleared all four of its pre-registered bars on the same 2,009 pairs: chrF2 58.96 → 60.00, BLEU 29.57 → 30.73 (a paired bootstrap over sentences puts the gains at +1.04 [+0.70, +1.35] chrF2 and +1.16 [+0.70, +1.62] BLEU, 0 of 1,000 resamples at or below zero); the Bosnian form rate on 338 audited bench targets 94.3% → 99.2% (244 of 246 decided, training/RESULTS-en-bs-formrate.md); and the >>bos_Latn<< label still steers, its gap against >>hrv<< going 21.8 → 22.5 points. The served build was rebuilt with the adapter merged and published on 8 September; scripts/fetch_models.py pulls it as translator-en-bs/. Its base, tc-base, is a smaller model than the forward direction's, so the two directions are not of comparable quality.

Speech

200 held-out FLEURS Bosnian clips, the same clips in every column. "Today" is the listener the bundle ships, whisper-large-v3, shipped by the owner's decision and refused at its gate; the gated whisper-small stays beside it as the baseline.

	the first builds (unrecorded)	first recorded	today
Word error rate	> 55%	38.5% (stock Whisper-small, training/RESULTS-speech.md)	11.9% (whisper-large-v3, shipped by decision, refused at its gate; the gated whisper-small reads 34.9%)

The two listeners against each other on the same 200 clips, one scorer, one process (training/SPEECHBENCH-gate.txt, 7 September):

	whisper-small, the gated listener	whisper-large-v3, shipped
Word error	34.9%	11.9%
Bosnian term recall	60.0%	89.1% (+29.1, p = 0.0000)
Croatian form written where Bosnian was said	5.3%	6.5% (+1.2, p = 0.48; on all 925 clips 1.1% → 6.1%, p = 0.018)

The gated small's own fine-tune, stock to tuned on those clips under the earlier scorer: word error 38.5% → 34.9%, Bosnian term recall 65.9% → 68.2%, wrong-variety substitutions 5.1% → 3.3%.

The larger listener: refused at its gate, shipped by decision. A whisper-large-v3 fine-tune reads 11.9% word error against the gated whisper-small's 34.9% on the same 200 clips (training/SPEECHBENCH-gate.txt). It has to clear three rows, not one: word error, Bosnian term recall, and Croatian substitution — how often it writes the Croatian form of a word where the Bosnian one was said.

• Its gate, run 7 September, refused it by one word on the Croatian row.
• The pre-registered last look, all 925 clips and both listeners on 8 September (training/speech-instrument/), refused it again, and not by one word: Croatian substitution 1.1% → 6.1% (1 of 87 against 8 of 131 decided targets, p = 0.018), the same two words over and over (Europom for evropom, vjerojatno for vjerovatno). Word error and term recall passed by wide margins.
• On the project's own scale (training/RUBRIC.md, computed for the first time in that run) the two listeners read 39.5% and 14.1%: band 3 against band 8.
• By rule 3 of that pre-registration whisper-large-v3 is closed to any further look: no other split, normaliser or instrument. That closes the question of whether it passes. What ships is a separate decision.
• The owner's decision, 8 September, evening: it ships anyway. With both readings on the table, the owner chose the larger listener for what it gets right, 14.1% against 39.5% of words wrong and 72% against 50% of Bosnian-specific words recovered, and accepted what it gets wrong: two Croatian spellings, Europom and vjerojatno, in 8 of 131 decided targets. The gate's result stands as written, in training/PREREGISTRATION.md with the reason. The bundle carries this listener only under its fingerprint named on the publish command (--allow-listen e6bb58483586b06c), every fresh install is told it is not the gated listener, and the gated whisper-small (34.9%) stays beside it as the baseline every listener is measured against.
• How it got there, not tidied. The 4–5 September reader publish swept the Mac's models/lilly/listen, already large-v3, into Safak11/lilly before any gate had run. That is the failure the fail-stop rules exist to prevent. It was taken out at 11:31 UTC on 8 September and put back that evening by the decision above, with the numbers on the table this time.

Photographs

Two sets of Bosnian signs from Wikimedia Commons, each transcribed by two readers independently, seeing neither each other's work nor any model's guess; only words both of them saw are in the answer key. The 40 are the original set (373 agreed words). test-v2 is 280 photographs drawn from the same pool, 132 with text, 2,907 agreed words, never trained on by anything.

	the 40	test-v2 (132 photographs)
the first builds (unrecorded)	< 30% found, > 280 invented	—
first recorded — the first reader (training/RESULTS-ocr.md)	36.0% found, 224 invented	—
EasyOCR, stock	48.0% found, 188 invented	30.0% found
EasyOCR fine-tuned on real crops (the reader until 5 Sep 2026)	54.5% found, 182 invented	34.6% found, 2,071 invented
PaddleOCR PP-OCRv6, untrained, confidence floor 0.9 — the reader now	67.0% found, 65 invented	57.8% found, 450 invented

The confidence floor is there because without it PP-OCRv6 read 60.0% and invented 2,373. The engine was chosen by a rule written before the run (training/PREREGISTRATION.md), on the big set, not the small one: the 40 alone had said the fine-tuned reader read 54.7%, and the 132 say 34.6%.

Thresholds are written before the run

Deciding measurements and their pass marks live in training/PREREGISTRATION.md, fixed before any number exists. Two retraining arms were run for the translator and the pre-written tie-break chose the one with the lower headline BLEU, because the rule said chrF2 decides. Published scores are bound to the weights by content hash, so the numbers and the model cannot drift apart.

---

What it cannot do yet

This section exists because a README that only lists wins is not worth trusting.

• The Bosnian-specific claim is not proven for the forward direction. A benchmark of 346 cases built from terms that separate Bosnian from Croatian and Serbian moves 91.7% → 92.2% at p = 0.360. The base model is already trained across South Slavic and arrives at 91.7% on its own, so there is very little room above it. (The reply direction is the first place this claim can be tested head-on, and there it holds: 94.3% → 99.2%.)
• The gain is concentrated in news prose. Broken out by corpus, the fine-tuning is worth +3.05 BLEU on news text and −0.82 BLEU on talks. Nothing measured here separates learned better Bosnian from adapted to news style.
• Against an outside system Lilly wins, and mostly not on its own merit. facebook/nllb-200-distilled-600M on the same 2,009 FLORES-200 pairs scores 36.49 BLEU bs→en against Lilly's 42.14 (whole rows through training/evaluate.py, not the served path), and 26.07 en→bs against 29.57 — a model 2.6× and 8× larger, beaten in both directions (training/RESULTS-outside-baseline.md). But the untouched Helsinki base already accounts for +5.11 of that +5.65 BLEU. What the fine-tune itself adds is small and its sign depends on the path: −0.79 chrF2 on whole rows, +0.15 on the path the product serves (43.03 / 67.81 against a tag-stripped base at 41.77 / 67.66 on all 2,009 pairs, re-measured 8 September; training/RESULTS-product.md). Its clearest win is not in either column: 308 of 1,012 base outputs leaked the model's language tag into the text, and 0 do after. In the reply direction that row was measured on the base (29.57 BLEU / 58.96 chrF2), before the 8 September fine-tune (30.73 / 60.00), so that margin is the base's. The win belongs largely to OPUS-MT, which this project builds on and did not train. And NLLB-600M is the distilled small variant: Google, DeepL, the 3.3B NLLB and the large general models were not tested, so none of this is a claim about the state of the art.
• English in is unmeasured. Since 8 September the listener hears English and the reader reads English photographs, because Whisper is multilingual and PP-OCRv6 reads Latin script whatever the language — but no held-out English set has been scored here, so there is no number for either. And the voice that says the Bosnian answer is not Bosnian: Piper's sr_RS voice, Serbian phonemes over recordings its card attributes to the Sorbian Institute, with numbers spelled out in the Serbian form. Nobody has yet measured how a Bosnian speaker hears it. Through Lilly's own listener it is heard with 22.3% of words wrong on the 200-clip test prefix, against 11.7% for the human recordings — and a voice trained on the FLEURS recordings themselves (9 September, pre-registered, training/RESULTS-speak-bs.md) was heard at 53.9% and did not ship. The recordings that exist are not enough for a voice; a clean hour from a native speaker is the untested path.
• English → Bosnian stays the weaker direction. The fine-tune cleared its bars (above) and has been in the bundle since 8 September, but it starts from a smaller base than the forward direction and reads 60.00 chrF2 where the forward direction reads 68.10.
• The photograph scores are recognition, not phone reality. The evaluated images come from Wikimedia Commons. Real photographs taken on a phone in Bosnia would be the honest test, and there is not a labelled set of them yet. The Commons images are also downscaled: a sign whose Commons original is 3968 px wide is scored here at 1280 px, and the app reads at up to 2 MP, so these numbers, if anything, understate what the reader does on a full-resolution photo. Measuring that is pre-registered and not yet run.
• The photograph score is not yet a valid score by the project's own scale. training/RUBRIC.md refuses to grade the reader below 200 real photographs with text; test-v2 has 132. Another 160 photographs (test-v2b) are fetched and waiting for two blind transcriptions.
• One test set. FLORES is professionally translated and even in register. Real user input is not.

The full write-up of every limit is on the model card and in training/.

---

Training

Training runs on Kaggle GPUs. This Mac commits, launches, polls, and installs the result; it does not run multi-hour fine-tunes. The boundary is written down in docs/V2-BOUNDARIES.md.

Lane	What it does	Gate before anything ships
Translation	LoRA or full fine-tune, either direction → adapter zip	pre-registered bars on FLORES, form rate and label steering
Speech half 1	one epoch → lilly-listen-half1.zip	training exits 0; no quality claim here
Speech half 2	resumes for epoch 2, then scores WER → lilly-listen.zip	AFTER WER on the 200 held-out clips, never skipped; shipping is decided by the instrument below
Speech instrument	scores two listeners on all 925 clips	the three gate rows, both not either
OCR	harvest, real crops, synthetic crops → lilly-read.zip	install gate must pass; the line is paused, see the roadmap

A known failure stops the kernel. A COMPLETE status, a leftover zip from an earlier run, or hitting the 12-hour wall does not override a gate.

python3 scripts/preflight_kaggle.py
python3 scripts/kaggle_train.py speech          # half 1
python3 scripts/kaggle_train.py ocr             # in parallel if a GPU slot is free
python3 scripts/kaggle_train.py speech-half2    # only after half 1 is COMPLETE
python3 scripts/kaggle_train.py speech-instrument   # 925 clips, both listeners: the last look at large-v3
python3 scripts/kaggle_train.py translation-en-bs   # the reply direction, LoRA, pre-registered bars
python3 scripts/kaggle_train.py speak-bs            # a Bosnian voice from FLEURS (Piper, warm-started); refused, see RESULTS-speak-bs.md
python3 scripts/kaggle_train.py speak-parla         # a voice from ParlaSpeech-HR, served as the mean of its speakers, same judge
python3 scripts/kaggle_train.py outside-baseline    # NLLB-200 on the same FLORES pairs
python3 scripts/kaggle_poll.py                  # CANCEL or ERROR counts as failure

What gets trained next, and what does not, is in docs/V4-PLAN.md. The reader's queue and do-not-repeat list are in docs/OCR-ROADMAP.md. How to write a notebook that fails loudly: docs/kaggle-notebooks.md. The list of failures already paid for: docs/kaggle-fail-stop.md.

---

Repo map

Path	What is in it
app/	FastAPI server, the five abilities in both directions, the web UI
models/lilly/	Offline weights, model card, attribution notice
training/	Notebooks, training and evaluation scripts, every results file
bench/	The Bosnian-versus-neighbours benchmark and how its cases are built
scripts/	kaggle_train.py, preflight, poll, fetch, publish
docs/	Plans, boundaries, fail-stop rules, white paper — index in docs/README.md
data/	Corpora and OCR crops (large, usually local only)
space/	Hugging Face Space packaging

---

Status

• Translate, listen, speak, read, web app, correction pipeline
• Published weights and model card with every score and every limit
• Pre-registered thresholds and hash-bound results
• English → Bosnian fine-tune — all four pre-registered bars cleared 8 Sep (chrF2 +1.04, BLEU +1.16, form rate 99.2%, label gap 22.5); built and published 8 Sep
• Larger speech model — trained (11.9% word error); refused at the gate 7 Sep and at the pre-registered last look 8 Sep (Croatian 1.1% → 6.1%, p = 0.018); closed to further looks by rule 3; shipped 8 Sep evening by the owner's decision, refused row and all, under a fingerprint named on the publish command
• A valid photograph score: test-v2b's 160 photographs transcribed blind, then one score on the union
• Re-measure the served translation path after the ordinal splitter fix — done 8 Sep on Kaggle, both splitters on one T4: devtest 42.49 → 43.25 BLEU, 67.69 → 68.10 chrF2 (p = 0.001); device drift within noise
• A labelled set of real phone photographs from Bosnia

---

Credits

The weights come from Helsinki-NLP and the OPUS-MT project, OpenAI and SYSTRAN, JaidedAI and Clova AI Research, PaddlePaddle, and hexgrad. CTranslate2 and peft shape the build. Please credit them rather than this repository. Every license was checked against the project's own page and is listed in models/lilly/NOTICE.md. The OPUS-MT authors ask to be cited; the citation is on the model card.

Built by @ssaaffaakk.