open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_SimNPO_lr2e-05_b3.5_a1_d1_g0.125_ep10

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:1BQuant:BF16Ctx Length:32kPublished:May 24, 2025Architecture:Transformer Warm

The open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_SimNPO_lr2e-05_b3.5_a1_d1_g0.125_ep10 model is a 1 billion parameter instruction-tuned language model with a 32768 token context length. This model is specifically designed for unlearning, utilizing a SimNPO method to forget specific information. Its primary differentiation lies in its ability to selectively remove knowledge, making it suitable for privacy-preserving AI applications or adapting models to new data without retaining outdated information.

Loading preview...

Model Overview

This model, open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_SimNPO_lr2e-05_b3.5_a1_d1_g0.125_ep10, is a 1 billion parameter instruction-tuned language model. It features a substantial context length of 32768 tokens, allowing it to process and generate longer sequences of text.

Key Capabilities

  • Unlearning Mechanism: The model incorporates a SimNPO (Similarity-based Negative Preference Optimization) method, specifically trained to "forget" certain information. This is indicated by forget10 in its name, suggesting it has undergone a process to remove 10 specific data points or concepts.
  • Instruction Following: As an instruction-tuned model, it is designed to understand and execute commands or prompts given in natural language.
  • Large Context Window: The 32768 token context length enables the model to maintain coherence and draw information from extensive input texts, which is beneficial for complex tasks requiring broad understanding.

Good for

  • Privacy-Preserving AI: Ideal for scenarios where specific sensitive data needs to be removed from a trained model without retraining from scratch.
  • Model Adaptation: Useful for updating models to new regulations or datasets by selectively unlearning outdated or incorrect information.
  • Research in Machine Unlearning: Provides a practical example and baseline for exploring and developing advanced unlearning techniques in large language models.