nllb200-en-formosan-spm8k

Direction: English to Formosan
Base model: facebook/nllb-200-distilled-600M
Recipe: nllb200-spm8k-directional-v3
Release: 20260809-210523, validation-selected step 280,000

This is a directional model for 15 Formosan languages. It uses the private_no_bible leakage-controlled corpus, Formosan-aware 8k SentencePiece extension, balanced language/source sampling, and direction/domain/dialect control tags. The model weights are public, but the private training corpus is not included.

Model details

Item	Value
Base revision	f8d333a098d19b4fd9a8b18f94170487ad3f821d
Training rows	573,657
Effective batch size	64
Maximum sequence length	384
Learning rate	2e-05
Precision	bf16
Checkpoint selection	Human validation chrF2
Formosan text	kindOf=standard, formosan-mt-standard-v3
Corpus SHA-256	e3feeaf7c3c51b9cd5c4b0537ffe44a370e7d02723f34b467479b6c4f5f0ea77
Training profile SHA-256	34f45832bdedc1b8322b26936dd88667dfecb360d941a7f54f60ae31a1fb4e10

Usage

import torch
from transformers import AutoModelForSeq2SeqLM, NllbTokenizer


model_id = "FormosanBank/nllb200-en-formosan-spm8k"
tokenizer = NllbTokenizer.from_pretrained(model_id, use_fast=False)
model = AutoModelForSeq2SeqLM.from_pretrained(model_id)
model.to("cuda" if torch.cuda.is_available() else "cpu")
NLLB_LIDS = {'ami': 'ami_Latn', 'bnn': 'bnn_Latn', 'ckv': 'ckv_Latn', 'dru': 'dru_Latn', 'pwn': 'pwn_Latn', 'pyu': 'pyu_Latn', 'ssf': 'ssf_Latn', 'sxr': 'sxr_Latn', 'szy': 'szy_Latn', 'tao': 'tao_Latn', 'tay': 'tay_Latn', 'trv': 'trv_Latn', 'tsu': 'tsu_Latn', 'xnb': 'xnb_Latn', 'xsy': 'xsy_Latn'}

def translate(text, lang_code, source_bucket="unknown", dialect="default"):

    tokenizer.src_lang = 'eng_Latn'
    prompt = (
        f"<to_{lang_code}> <src_eng> <dom_{source_bucket}> "
        f"<dialect_{dialect}> {text}"
    )
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    output = model.generate(
        **inputs,
        decoder_start_token_id=tokenizer.eos_token_id,
        forced_bos_token_id=tokenizer.convert_tokens_to_ids(NLLB_LIDS[lang_code]),
        max_new_tokens=256,
        num_beams=4,
    )
    return tokenizer.batch_decode(output, skip_special_tokens=True)[0]

print(translate('He went home.', "ami"))

The control tags are part of the training contract. Use unknown and default when source bucket or dialect metadata is unavailable.

Evaluation

The best checkpoint was selected on human validation chrF2. Test references are human sentence pairs; synthetic pivots and lexical entries are train-only. The headline result uses default metadata controls, so it does not assume access to test-set domain or dialect labels.

Split	Rows
Train	573,657
Test	63,140
Validate	20,733

Scope	BLEU	chrF2	TER
Hard test	7.14	30.47	83.88
Selection validation	7.98	33.96	79.31

Test empty-output rate: 0.1615%.

Confidence intervals

Stratified bootstrap, 200 samples, 95% confidence.

Metric	Lower	Upper
BLEU	6.98	7.28
chrF2	30.35	30.64
TER	83.17	84.69

Language	Samples	BLEU	chrF2	TER
ami	11,017	5.43	27.77	84.23
bnn	6,049	3.07	29.87	84.86
ckv	2,910	15.06	39.37	72.42
dru	5,846	0.53	16.19	110.56
pwn	5,811	8.62	33.68	81.01
pyu	4,315	11.62	33.68	86.56
ssf	1,630	17.39	45.69	70.56
sxr	2,081	7.38	43.74	78.08
szy	1,852	11.13	38.22	74.16
tao	2,257	9.31	35.38	82.27
tay	6,411	2.55	21.22	94.70
trv	6,502	12.27	32.59	70.59
tsu	2,003	1.28	22.25	85.93
xnb	2,682	2.81	33.33	89.59
xsy	1,774	21.09	44.75	67.02

The corpus gate enforces standard-tier Formosan text, at least 7.5% test and 2.5% validation per language, human sentence-only evaluation, and zero exact, skeleton, one-edit, configured high character n-gram, or document train/evaluation conflicts. This release passed all gates: exact 0, skeleton 0, one-edit 0, character n-gram 0, and document 0.

See eval/metrics.json for sacreBLEU signatures, per-language, source, dialect, and length diagnostics. publication.json records the corpus, profile, run, and checkpoint hashes used for this release.

Intended use

This model supports research, corpus development, and assisted translation for the 15 included Formosan languages. It is designed for the exact prompt and generation contract shown above.

Limitations

Outputs require knowledgeable speaker review. Aggregate metrics hide large differences among languages and domains. This model is not suitable for authoritative, medical, legal, or safety-critical translation.