Mushari440/Qwen3-8B-SFT-v2
Mushari440/Qwen3-8B-SFT-v2 is a supervised fine-tuned (SFT) causal language model developed by Mushari Alothman, based on Qwen3-8B-Base. This model is specifically optimized for accurate and clean instruction following across both Arabic and English tasks. It excels in applications such as Arabic and English MCQ answering, context-based QA/RAG, and general instruction following, leveraging curated datasets for its fine-tuning.
Loading preview...
Model Overview
Mushari440/Qwen3-8B-SFT-v2 is a supervised fine-tuned (SFT) causal language model developed by Mushari Alothman. It is built upon the Qwen3-8B-Base architecture and is specifically optimized for robust performance in both Arabic and English language tasks. The model operates under an Apache 2.0 license.
Key Capabilities
- Bilingual Proficiency: Optimized for accurate instruction following in both Arabic and English.
- Question Answering: Designed for effective Arabic and English Multiple Choice Question (MCQ) answering.
- Contextual Understanding: Strong capabilities in context-based Question Answering (QA) and Retrieval-Augmented Generation (RAG).
- General Instruction Following: Capable of adhering to a wide range of general instructions.
Training Details
The model underwent Supervised Fine-Tuning (SFT) using bf16 mixed precision. Its training data comprises curated Arabic and English datasets, specifically focusing on:
- MCQ tasks
- QA / RAG / context understanding scenarios
- General instruction-following data
Intended Use Cases
This model is well-suited for direct applications requiring accurate instruction following and contextual understanding in Arabic and English. It is not intended for safety-critical or real-time decision-making, nor for generating factual guarantees without external verification.