open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_RMU_lr1e-05_layer5_scoeff100_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_RMU_lr1e-05_layer5_scoeff100_epoch10 model is a 1 billion parameter instruction-tuned language model based on the Llama-3.2 architecture. This model is specifically designed for "unlearning" specific information, indicated by the "forget10" and "RMU" (Retain-Memory Unlearning) in its name. It is likely optimized for scenarios requiring selective forgetting or privacy-preserving modifications of pre-trained knowledge. Its primary use case involves research and applications in machine unlearning and model editing.

Loading preview...

Model Overview

This model, open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_RMU_lr1e-05_layer5_scoeff100_epoch10, is a 1 billion parameter instruction-tuned language model built upon the Llama-3.2 architecture. Its naming convention, particularly "forget10" and "RMU" (Retain-Memory Unlearning), strongly suggests its focus on the emerging field of machine unlearning. This involves the selective removal or modification of specific information from a pre-trained model without requiring a full retraining process.

Key Characteristics

  • Architecture: Llama-3.2-1B-Instruct base model.
  • Parameter Count: 1 billion parameters.
  • Context Length: 32768 tokens.
  • Specialization: Designed for machine unlearning, likely focusing on techniques to make a model "forget" specific data points or patterns.
  • Instruction-Tuned: Capable of following instructions, which is typical for conversational or task-oriented LLMs.

Potential Use Cases

  • Research in Machine Unlearning: Exploring and developing new methods for removing data from trained models.
  • Privacy-Preserving AI: Investigating ways to mitigate privacy risks by selectively erasing sensitive information.
  • Model Editing: Modifying model behavior or knowledge without extensive retraining.
  • Ethical AI Development: Addressing biases or undesirable behaviors by unlearning specific patterns.