You can deploy this model and start earning money today!

jizhiovo/
Hy-MT2-1.8B-mlx-4bit

Name: Hy-MT2-1.8B-mlx-4bit
Availability: InStock

Views

No views yet

Hy-MT2-1.8B-mlx-4bit

4-bit MLX quantization of Tencent's tencent/Hy-MT2-1.8B multilingual translation model, for fast on-device inference on Apple Silicon (native Metal).

Base model: tencent/Hy-MT2-1.8B (Apache-2.0)
Quantization: 4-bit (≈4.5 bits/weight), via mlx_lm.convert
Size: ~974 MB on disk, ~1.1 GB resident
Languages: 33 (Mandarin, Cantonese, English, French, Japanese, Korean, Arabic, Russian, …)
Converted with mlx-lm 0.31.3.

This is a derivative redistribution permitted under the base model's Apache-2.0 license. All credit for the model goes to Tencent Hunyuan.

Use with mlx-lm

bash

pip install mlx-lm

# one-shot
python -m mlx_lm generate --model jizhiovo/Hy-MT2-1.8B-mlx-4bit \
  --prompt "Translate into Chinese.\n\nThe quick brown fox jumps over the lazy dog."

# OpenAI-compatible server
python -m mlx_lm server --model jizhiovo/Hy-MT2-1.8B-mlx-4bit --host 127.0.0.1 --port 8110

Reproduce the conversion

bash

python -m mlx_lm convert --hf-path tencent/Hy-MT2-1.8B -q --q-bits 4 \
  --mlx-path Hy-MT2-1.8B-mlx-4bit