StudyAbroadGPT-7B-LoRa-Kaggle

StudyAbroadGPT is a fine-tuned LoRA adapter based on Mistral-7B-Instruct-v0.3. It is designed to provide accurate, structured, and context-aware guidance for students pursuing education abroad. The model specializes in answering queries regarding university applications, scholarships, visa regulations, and accommodation.

This version was trained using the Unsloth library on a Tesla T4 GPU (via Kaggle), demonstrating the feasibility of fine-tuning LLMs on resource-constrained hardware.

📊 Model Details

• Base Model: unsloth/mistral-7b-instruct-v0.3-bnb-4bit
• Fine-Tuning Method: LoRA (Low-Rank Adaptation)
• Quantization: 4-bit NF4 (via Unsloth)
• Framework: Unsloth
• Developer: MD Millat Hosen (Sharda University)

📂 Dataset

The model was trained on the StudyAbroadGPT-Dataset, a synthetic dataset containing 2,274 high-quality student-advisor conversation pairs generated via Gemini Pro.

• Dataset Link: millat/StudyAbroadGPT-Dataset

🚀 Usage

You can easily run this model using the unsloth library for faster inference.

Installation

pip install "unsloth[colab-new] @ git+[https://github.com/unslothai/unsloth.git](https://github.com/unslothai/unsloth.git)"
pip install --no-deps "xformers<0.0.27" "trl<0.9.0" peft accelerate bitsandbytes