open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_UNDIAL_lr1e-05_beta10_alpha1_epoch10

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:1BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:May 15, 2025Architecture:Transformer Warm

The open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_UNDIAL_lr1e-05_beta10_alpha1_epoch10 model is a 1 billion parameter instruction-tuned language model. This model is part of the Llama-3.2 family and is specifically designed for unlearning tasks, indicated by its 'unlearn_tofu' and 'forget10_UNDIAL' naming conventions. Its primary differentiator lies in its focus on selective forgetting, making it suitable for research and applications requiring the removal of specific information from a pre-trained model.

Loading preview...

Model Overview

This model, open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_UNDIAL_lr1e-05_beta10_alpha1_epoch10, is a 1 billion parameter instruction-tuned language model based on the Llama-3.2 architecture. It is specifically developed for unlearning tasks, aiming to selectively remove or 'forget' certain information that the model was previously trained on. The naming convention, including 'unlearn_tofu' and 'forget10_UNDIAL', highlights its specialization in this area.

Key Characteristics

  • Architecture: Llama-3.2-1B-Instruct, indicating a 1 billion parameter model fine-tuned for instruction following.
  • Specialization: Designed for model unlearning, focusing on the selective removal of specific data or knowledge.
  • Context Length: Supports a context length of 32768 tokens.

Potential Use Cases

  • Research in Machine Unlearning: Ideal for exploring and evaluating different unlearning algorithms and their effectiveness.
  • Data Privacy Compliance: Could be used in scenarios requiring the removal of sensitive or proprietary information from a deployed model.
  • Model Debugging: Useful for investigating how specific training data influences model behavior by selectively removing its influence.