ClaudioSavelli/FAME_FT_llama32-1b-2p5-instruct-qa

Model Overview

ClaudioSavelli/FAME_FT_llama32-1b-2p5-instruct-qa is a 1 billion parameter instruction-tuned model, built upon the meta-llama/Llama-3.2-1b-Instruct base. Its primary distinction lies in its fine-tuning methodology, which incorporates an "unlearning" technique specifically adapted for the FAME (Fine-tuning method for the FAME setting) framework. This approach aims to modify the model's behavior or knowledge post-training.

Key Characteristics

• Base Model: Utilizes the meta-llama/Llama-3.2-1b-Instruct architecture.
• Parameter Count: Features 1 billion parameters, offering a balance between performance and computational efficiency.
• Context Length: Supports a substantial context window of 32768 tokens, enabling processing of longer inputs and maintaining conversational coherence over extended interactions.
• Fine-tuning Method: Employs a specialized fine-tuning method focused on "unlearning" within the FAME setting, as detailed in the associated research paper.

Intended Use Cases

This model is particularly suited for research and applications involving:

• Question Answering (QA): Designed for instruction-based question-answering tasks.
• Model Unlearning Research: Ideal for exploring and implementing techniques for removing specific information or behaviors from pre-trained language models.
• Controlled Information Generation: Useful in scenarios where precise control over the model's knowledge base is required, especially in the context of FAME settings.

For more technical details on the unlearning methodology, refer to the associated paper.