Hunyuan-MT2 1.8B — 6-bit (MLX)

A 6-bit, group-size-64 MLX quantization of Tencent's Hy-MT2-1.8B, for on-device translation on Apple Silicon. ~33 languages, particularly strong into Chinese / Japanese / Korean. Architecture unchanged — this repo only re-quantizes the original weights.


Base model	`tencent/Hy-MT2-1.8B`
Quantization	6-bit, group size 64, affine
Format	MLX `safetensors`
Size	~1.45 GB

License & attribution

Subject to the Tencent Hunyuan Community License Agreement — see the license. Use restrictions apply (including commercial-use conditions, and a prohibition on using outputs to train other models). Powered by Tencent Hunyuan. Credit: Tencent. This repository only provides an MLX-quantized copy of the released weights and is redistributed under the same license.

Downloads last month: 73

Safetensors

Model size

0.4B params

Tensor type

BF16

U32

MLX

Hardware compatibility

6-bit

Model tree for huggingfinger0/Hunyuan-MT2-1.8B-6bit-MLX

Base model

tencent/Hy-MT2-1.8B

Quantized

(28)

this model