Pith. sign in

REVIEW 2 cited by

Recovering the Pre-Fine-Tuning Weights of Generative Models

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2402.10208 v2 pith:754W2V7D submitted 2024-02-15 cs.LG cs.CLcs.CRcs.CV

classification cs.LGcs.CLcs.CRcs.CV
keywords pre-fine-tuningrecoverweightsmethodmodelmodelsgenerativelarge-scale
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

The dominant paradigm in generative modeling consists of two steps: i) pre-training on a large-scale but unsafe dataset, ii) aligning the pre-trained model with human values via fine-tuning. This practice is considered safe, as no current method can recover the unsafe, pre-fine-tuning model weights. In this paper, we demonstrate that this assumption is often false. Concretely, we present Spectral DeTuning, a method that can recover the weights of the pre-fine-tuning model using a few low-rank (LoRA) fine-tuned models. In contrast to previous attacks that attempt to recover pre-fine-tuning capabilities, our method aims to recover the exact pre-fine-tuning weights. Our approach exploits this new vulnerability against large-scale models such as a personalized Stable Diffusion and an aligned Mistral.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Activation Approximations Can Incur Safety Vulnerabilities Even in Aligned LLMs: Comprehensive Analysis and Defense

    cs.CR 2025-02 conditional novelty 7.0 of 10

    Activation approximations used for faster LLM inference can dramatically increase jailbreak success rates on aligned models, and the paper's QuadA training method restores robust refusal.

  2. Task-Specific Adaptation with Restricted Model Access

    cs.CV 2025-02 conditional novelty 5.0 of 10

    A new gray-box fine-tuning framework adapts frozen, hidden foundation models using only gradient access, with input/output adapters (DGA) or intermediate tokens (LGA), and matches full-access methods on several retrie...

Pith tools