Mirror of PrivateWhisper/refine-qwen3-1.7b-6bit at revision fe5cd8d3fd45cb3594bbd6b7b16f918a62d2ce21, kept by the PrivateWispr team as a backup host. Licence: not declared upstream; the model is a LoRA-fine-tuned, 6-bit MLX quantization of Qwen3-1.7B (Apache-2.0), and the Qwen3 licence applies. Original authors: the PrivateWispr team (the model was trained for PrivateWispr, a local dictation app for macOS).

The weights, tokenizer and config files in this repo are byte-identical to the upstream revision above. Only this README was added.

Downloads last month: 11

Safetensors

Model size

2B params

Tensor type

U32

BF16

MLX

Hardware compatibility

6-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for codeclawd/refine-qwen3-1.7b-6bit

Base model

Qwen/Qwen3-1.7B-Base

Finetuned

Qwen/Qwen3-1.7B

Quantized

(420)

this model