FADE TOFU Checkpoints
Collection
Checkpoints for reproducing FADE on TOFU (arXiv:2510.12981): full model, retain-only oracles, and unlearned LoRA checkpoints. • 7 items • Updated
fade-tofu-full
Llama 3.1 8B fine-tuned on the complete TOFU dataset (5 epochs, lr 1e-5) -- the starting-point checkpoint every unlearning method in this collection was trained from, and one of FADE's two comparison points (against the retain-only oracles) before any unlearning is applied.
Part of the checkpoint collection for reproducing FADE on TOFU (arXiv:2510.12981, "Reference-Specific Unlearning Metrics Can Hide the Truth: A Reality Check"). Code, training/eval scripts, and full documentation: https://github.com/sc782/fade_unlearning/tree/main/fade-tofu
Notice
Llama 3.1 is licensed under the Llama 3.1 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.
Built with Llama.
Use of this model is governed by the Llama 3.1 Community License Agreement and Meta's Acceptable Use Policy, in addition to any terms in this repository.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for sungjuncho/Llama-3.1-8B-fade-tofu-full
Base model
NousResearch/Meta-Llama-3.1-8B-Instruct