open-unlearning/neg_tofu_Llama-3.2-1B-Instruct_retain90_lr2e-05_wd0.01_epoch10
The open-unlearning/neg_tofu_Llama-3.2-1B-Instruct_retain90_lr2e-05_wd0.01_epoch10 is a 1 billion parameter instruction-tuned language model, likely based on the Llama 3.2 architecture, with a 32768 token context length. This model is specifically designed for open unlearning research, focusing on retaining 90% of original knowledge while unlearning specific negative information. Its primary use case is in exploring and evaluating methods for targeted knowledge removal in large language models.
Loading preview...
Model Overview
This model, open-unlearning/neg_tofu_Llama-3.2-1B-Instruct_retain90_lr2e-05_wd0.01_epoch10, is a 1 billion parameter instruction-tuned language model. It is characterized by its focus on open unlearning, a research area dedicated to selectively removing specific information from a model while preserving its general capabilities. The model is likely built upon the Llama 3.2 architecture and supports a substantial context length of 32768 tokens.
Key Capabilities
- Targeted Knowledge Unlearning: Designed to demonstrate and evaluate the removal of specific negative information.
- Knowledge Retention: Aims to retain approximately 90% of its original knowledge base during the unlearning process.
- Instruction Following: As an instruction-tuned model, it can process and respond to user prompts effectively.
- Large Context Window: Benefits from a 32768-token context length, allowing for processing longer inputs and maintaining conversational coherence over extended interactions.
Good For
- Research in Open Unlearning: Ideal for researchers and practitioners exploring techniques for model editing and knowledge removal.
- Evaluating Unlearning Methods: Can be used as a benchmark or testbed for new unlearning algorithms.
- Understanding Model Behavior: Provides insights into how models adapt and change when specific data is unlearned.
- Developing Safer AI Systems: Contributes to the development of models that can be updated to remove undesirable or harmful information.