Lingua Imperialis - NLLB-200 models (CTranslate2, int8)

CTranslate2 int8 conversions of Meta's NLLB-200 translation models, used by Lingua Imperialis, a free Warhammer 40,000: Darktide mod that translates in-game chat locally on the player's machine.

These are format conversions only. No retraining, no fine-tuning, no change to model behaviour.

Contents

Folder Base model model.bin Notes
model/ facebook/nllb-200-distilled-600M 622,596,105 B default; faster, lower memory
model-large/ facebook/nllb-200-1.3B 1,381,827,201 B higher accuracy

Each folder contains the four files the CTranslate2 runtime needs: model.bin, config.json, shared_vocabulary.json, sentencepiece.bpe.model.

The mod pins the SHA-256 of each model.bin and refuses to load a model that does not match, so these files must not be modified in place.

How they were produced

ct2-transformers-converter \
  --model facebook/nllb-200-distilled-600M \
  --quantization int8 \
  --copy_files tokenizer.json sentencepiece.bpe.model

(and the same for facebook/nllb-200-1.3B).

Licence and attribution

Licensed CC-BY-NC-4.0, inherited from the base models. NLLB-200 is a work by Meta AI; all credit for the models belongs to them. Non-commercial use only, consistent with the original licence. See the NLLB paper: No Language Left Behind: Scaling Human-Centered Machine Translation (NLLB Team et al., 2022).

Conversion and packaging by Wobin.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Wobin/lingua-imperialis-models

Finetuned
(35)
this model