Model Overview

This model, open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_SimNPO_lr5e-05_b3.5_a1_d1_g0.25_ep5, is a 1 billion parameter instruction-tuned language model built upon the Llama-3.2 architecture. It has a context length of 32768 tokens. The model's core distinction lies in its application of machine unlearning techniques, specifically using the SimNPO method to forget 10 designated items.

Key Capabilities

• Targeted Unlearning: Designed to remove specific information (10 items) from its knowledge base.
• Instruction Following: Retains general instruction-following abilities post-unlearning.
• Research in Unlearning: Serves as a valuable tool for exploring and evaluating machine unlearning methodologies.

Good For

• Machine Unlearning Research: Ideal for experiments and studies on how to effectively remove data from LLMs.
• Privacy-Preserving AI: Useful for developing models that can selectively forget sensitive or outdated information.
• Model Auditing: Can be used to test the efficacy of unlearning algorithms and their impact on model performance.