# 4-bit Quantization

This is a 4-bit quantized version converted from the Qwen3-30B model, suitable for text generation tasks on the MLX framework.

MedGemma-27B-Text-IT-4bit is an MLX-format model converted from Google's MedGemma-27B-Text-IT model, specifically optimized for medical and clinical reasoning tasks.

Moondream is a lightweight vision-language model designed for efficient cross-platform deployment. The 4-bit quantized version released on April 14, 2025 significantly reduces memory usage while maintaining high accuracy.

A Turkish law-specific language model based on Qwen3-14B, fine-tuned using the LoRA method and employing 4-bit quantization technology

A 3B parameter instruction fine-tuned model based on Qwen2.5 architecture, optimized for training speed using Unsloth and Huggingface TRL library

A roleplay dialogue model fine-tuned based on Qwen3-4B, excelling in generating coherent dialogues that align with character traits

This model is a fine-tuned version optimized from Mistral-7B-Instruct-v0.3, specifically designed for Q&A tasks in the field of forensic investigations, supporting advanced forensic reasoning and rapid knowledge retrieval.

UI-TARS-1.5-7B-4bit is a multimodal model focused on image-text-to-text conversion tasks, supporting the English language.

VL-Rethinker-72B-4bit is a multimodal model based on Qwen2.5-VL-7B-Instruct, supporting visual question answering tasks, and has been converted to MLX format for efficient operation on Apple devices.

Gemma 3 27B IT QAT 4bit is an MLX-format model converted from Google's original model, supporting image-to-text tasks.

Fine-tuned version based on Qwen2.5-VL-3B model, trained using Unsloth and Huggingface TRL library, achieving 2x inference speed improvement

A 4-bit GPTQ quantized version of the Qwen2.5-Omni-7B model, supporting multilingual and multimodal tasks.

This is a 4-bit quantized version converted from the mlabonne/gemma-3-27b-it-abliterated model, optimized for the MLX framework.

Llama model trained with Unsloth and Huggingface TRL library, achieving 2x inference speed improvement

Vision instruction fine-tuned model optimized with Unsloth, supporting multimodal task processing

4-bit quantized version based on Meta-Llama-3.1-8B-Instruct, efficiently trained using Unsloth and TRL libraries

This is the 4-bit quantized version of Qwen2-Audio-7B-Instruct, developed based on Alibaba Cloud's original Qwen model. It is an audio-text multimodal large language model.

A 4-bit quantized version based on Meta-Llama-3.1-8B-Instruct, trained using Unsloth and Huggingface TRL libraries, achieving 2x speed improvement.

A specialized summarization model fine-tuned for 2 rounds based on Qwen2-1.5B-Instruct

Omost's instruction fine-tuned model based on Llama3-8B, pre-trained with the Dolphin-2.9 dataset and quantized in 4-bit NF4 format.

A 4-bit quantized version of the LLAMA-3-8B model fine-tuned for medical Q&A

A text generation model fine-tuned based on unsloth/llama-3-8b-bnb-4bit, trained using Unsloth and TRL libraries with 2x speed improvement

This is a text generation model fine-tuned using Unsloth and TRL libraries on the Llama-3-70B model, achieving 2x faster training speed.

CogVLM is a powerful open-source vision-language model supporting tasks like object detection and visual question answering, featuring 4-bit precision quantization.

A fine-tuned language model based on the 4-bit quantized version of TinyLLaMA, efficiently trained using Unsloth and TRL libraries

InternLM-XComposer2 is a vision-language large model (VLLM) based on InternLM2, featuring advanced image-text understanding and creation capabilities.

A vision-language large model based on InternLM2, with outstanding image-text understanding and creation capabilities

Meditron 7B is a large language model in the medical field developed by the EPFL LLM Team. It is further pre-trained based on Llama-2-7B and focuses on medical knowledge encoding and clinical decision support.

A LoRA adapter fine-tuned on the Meta Llama 2 7B model, specifically designed for French-to-English text translation tasks.

A 4-bit GPTQ quantized model based on Pygmalion-6B, suitable for dialogue generation tasks, supporting English text generation