anmoldhandhania93/ANMOLGPT-4B-v0.2
ANMOLGPT-4B-v0.2 is a 4.5 billion parameter instruction-following language model developed by anmoldhandhania93, fine-tuned from Qwen3.5-4B using QLoRA with Unsloth Studio. This model aims to improve reasoning, instruction following, and overall language understanding. It is suitable for general-purpose text generation, conversational AI, and educational applications.
Loading preview...
ANMOLGPT-4B-v0.2 Overview
ANMOLGPT-4B-v0.2 is an instruction-following large language model, building upon the Qwen3.5-4B base model. Developed by anmoldhandhania93, it was fine-tuned using QLoRA with Unsloth Studio, focusing on enhancing reasoning, instruction following, and general language understanding. This release is the second public iteration in the ANMOLGPT family, emphasizing reproducible training and transparent evaluation.
Key Capabilities & Features
- Base Model: Fine-tuned from Qwen3.5-4B, a capable foundation model.
- Fine-tuning Method: Utilizes QLoRA for efficient fine-tuning with 4-bit quantization.
- Training Data: Trained on the Databricks Dolly 25K dataset.
- Performance: Shows comparable or slightly improved benchmark results over its predecessor, ANMOLGPT-3B-v0.1, across metrics like HellaSwag, PIQA, ARC-Easy, Winogrande, TruthfulQA MC2, and MMLU.
- Open-Source Focus: Part of a project dedicated to exploring efficient and transparent open-source LLM development.
Intended Uses
ANMOLGPT-4B-v0.2 is suitable for a variety of applications, including:
- Conversational AI and general-purpose text generation.
- Learning, experimentation, and prompt engineering.
- Educational applications and software development assistance.
Limitations
As an experimental open-source release, it may produce inaccurate or fabricated information and is not intended for safety-critical or production use without further validation. Future versions aim for comprehensive benchmarking and advanced capabilities like RAG and tool calling.