AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

AntiDote: Bi-level Adversarial Training for Tamper-Resistant LLMs

Debdeep Sanyal, Manodeep Ray, Murari Mandal

AAAI 2026 · Large language models › Instruction and fine-tuning · approach: Safety alignment after fine-tuning · 4 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add state-aware hypernetwork input (live activations instead of static prompt embedding) (ablation)AntiDote · Llama-3.2-3B harmful fine-tuningHarmful Score: 33.1 -> 5.7Improved
add state-aware hypernetwork input (ablation)AntiDote · Gemma-3-27B harmful fine-tuningHarmful Score: 34.8 -> 9.9Improved
add state-aware hypernetwork input (ablation)AntiDote · Gemma-3-12B harmful fine-tuningHarmful Score: 31.5 -> 9.1Improved
add state-aware input (ablation)AntiDote · Llama-3.2-3BFine-tune accuracy: 90.3 -> 90.8No clear effect
decouple capability loss onto the clean model (ablation)AntiDote · Llama-3.2-3BFine-tune accuracy: 85.4 -> 90.8Improved
decouple capability loss onto the clean model (ablation)AntiDote · Gemma-3-27BFine-tune accuracy: 90.2 -> 95.3Improved
decouple capability loss onto the clean model (ablation)AntiDote · Gemma-3-12BHarmful Score: 6.4 -> 5.9No clear effect
use AntiDote instead of BoosterBooster · Gemma-3-27B harmful fine-tuningHarmful Score: 13.5 -> 9.8Improved
use AntiDote instead of SFTSFT · Gemma-3-27B harmful fine-tuningHarmful Score: 44.2 -> 9.8Improved
use AntiDote instead of Booster on GSM8KBooster · Benign fine-tuning on GSM8KHarmful Score: 6.5 -> 8.8Hurt
use RMU unlearning instead of SFTSFT · Benign fine-tuning, 4 benchmarks averageFine-tune accuracy: FA 66.2 -> 64.2; HS 15.2 -> 12.2Mixed
use AntiDote instead of Booster in memory (12B model)Booster · Gemma-3-12BGPU memory (GB): 82.5 -> 73.6Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy