ClaudioSavelli/FAME_GD_llama32-1b-5-instruct-qa
ClaudioSavelli/FAME_GD_llama32-1b-5-instruct-qa is a 1 billion parameter language model derived from Meta's Llama-3.2-1b-Instruct, featuring a 32768 token context length. This model has been specifically 'unlearned' using the Gradient Difference method within the FAME (Forgetting in AI Models Explicitly) setting. It is designed for research into model unlearning techniques, particularly for scenarios requiring selective forgetting of learned information.
Loading preview...
Model Overview
ClaudioSavelli/FAME_GD_llama32-1b-5-instruct-qa is a 1 billion parameter instruction-tuned model, built upon the meta-llama/Llama-3.2-1b-Instruct architecture. Its primary distinction lies in its development using the Gradient Difference method for model unlearning within the FAME (Forgetting in AI Models Explicitly) setting. This process aims to selectively remove specific learned information from the model.
Key Capabilities
- Model Unlearning Research: Specifically designed for exploring and evaluating techniques to 'unlearn' or selectively forget data from large language models.
- Gradient Difference Method: Implements a particular unlearning approach, making it suitable for studies comparing different unlearning methodologies.
- Base Model Performance: Inherits the foundational capabilities of the Llama-3.2-1b-Instruct model prior to the unlearning process.
Good For
- Researchers and developers investigating model unlearning and data privacy in AI.
- Experiments requiring a model that has undergone a specific unlearning procedure (Gradient Difference).
- Understanding the impact of unlearning techniques on model performance and behavior.
For more technical details on the unlearning methodology, refer to the associated research paper.