You can deploy this model and start earning money today!

translate-studio/
NLLB-200-distilled-1.3B-int5-MLX

Name: NLLB-200-distilled-1.3B-int5-MLX
Availability: InStock

facebook/nllb-200-distilled-1.3B

quantized

cc-by-nc-4.0

Views

No views yet

NLLB-200 distilled 1.3B — int5 (MLX)

A 5-bit, group-size-64 MLX quantization of facebook/nllb-200-distilled-1.3B, for on-device translation on Apple Silicon. 200 languages.

Custom format. This is a purpose-built MLX seq2seq checkpoint (weights.safetensors + mlx_config.json + tokenizer/), not a standard mlx_lm / transformers checkpoint — it is loaded by a hand-written MLX encoder–decoder, not mlx_lm.load.


Base model	`facebook/nllb-200-distilled-1.3B`
Quantization	5-bit, group size 64
Size	~0.95 GB

License & attribution

CC-BY-NC-4.0, inherited from the base model — non-commercial use only, with attribution. Credit: the NLLB Team, Meta AI — see No Language Left Behind. This repository only provides an MLX-quantized copy of the released weights.

translate-studio/NLLB-200-distilled-1.3B-int5-MLX

NLLB-200 distilled 1.3B — int5 (MLX)

License & attribution

translate-studio/
NLLB-200-distilled-1.3B-int5-MLX