JoaoBoer/tofu_Llama-3.2-1B-Instruct_forget01_NPO

TEXT GENERATIONPricing:Input $0.108 / Output $0.804Concurrent Unit Cost:1Model Size:1BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Sep 10, 2026License:llama3.2Architecture:Transformer Featherless Exclusive Cold

The JoaoBoer/tofu_Llama-3.2-1B-Instruct_forget01_NPO is a 1 billion parameter instruction-tuned Llama-3.2 model developed by JoaoBoer, specifically unlearned on the TOFU 'forget01' split using the NPO method. This model serves as a weight-unlearning baseline within the Speculative-Decoding-Unlearning project. It is characterized by its targeted unlearning capabilities, making it suitable for research into model privacy and data removal techniques.

Loading preview...

Model Overview

JoaoBoer/tofu_Llama-3.2-1B-Instruct_forget01_NPO is a 1 billion parameter instruction-tuned Llama-3.2 model that has undergone a specific unlearning process. Developed by JoaoBoer, this model was unlearned on the TOFU forget01 dataset split using the NPO (Negative Preference Optimization) method, as part of the open-unlearning framework.

Key Characteristics

  • Targeted Unlearning: Specifically designed to forget information from the TOFU forget01 split.
  • NPO Method: Utilizes Negative Preference Optimization for the unlearning process.
  • Research Baseline: Serves as a weight-unlearning baseline or draft model in the Speculative-Decoding-Unlearning project.
  • Evaluation Metrics: Achieves an exact_memorization of 0.6310 and a forget_quality of 0.5786, with a model_utility of 0.5467, indicating its performance post-unlearning.

Use Cases

This model is particularly relevant for:

  • Research in Machine Unlearning: Investigating methods and effectiveness of removing specific data from trained models.
  • Privacy-Preserving AI: Exploring techniques to enhance model privacy by forgetting sensitive information.
  • Comparative Studies: Benchmarking against other unlearning techniques or serving as a component in speculative decoding research.

Finetunes
1 models