Modelpublic

morpknight/qwen3.5-4b-indonesian-legal-mlx

sourceHugging Faceapache-2.0updated 1mo agoView on Hugging Face

0likes93downloads

Model Card

Qwen3.5 4B Indonesian Legal — MLX BF16

BF16 MLX export of the Qwen3.5 4B Base model fine-tuned with the Indonesian legal-domain PEFT LoRA adapter.

Usage

bash

pip install mlx mlx-lm

mlx_lm.generate \
  --model morpknight/qwen3.5-4b-indonesian-legal-mlx \
  --prompt "Apa tujuan Undang-Undang Perlindungan Data Pribadi di Indonesia?" \
  --max-tokens 256 \
  --temp 0 \
  --extra-eos-token '<|im_end|>' \
  --chat-template-config '{"enable_thinking": false}' \
  --verbose False

The model is intended for Indonesian legal-domain assistance. It is not a substitute for review by a qualified legal professional, and answers should be checked against current official sources.

Conversion

The PEFT LoRA adapter was merged into the base model in BF16 and then exported with mlx_lm.convert. The MLX Qwen3.5 implementation used here is text-only; the vision tower is not included in this export.

See the project conversion manifest for provenance and validation details.