AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

Tree-based Dialogue Reinforced Policy Optimization for Red-Teaming Attacks

Ruohao Guo, Afshin Oroojlooy, Roshan Sridhar, Miguel Ballesteros, Alan Ritter, Dan Roth

ICLR 2026 · Adversarial robustness and security › Attacks on text and graphs · approach: Adversarial prompts for language models · 6 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 RL stage after SFT (ablation)DialTree attacker · HarmBench, closed-source avgASR (%): 29.6 -> 82.7Improved
add RL stage after SFT (ablation)DialTree attacker · HarmBench, open-source avgASR (%): 44.3 -> 80.2Improved
add format pruning (ablation)DialTree · HarmBench (Llama-3.1-8B)ASR (%): 62.5 -> 71.5Improved
add topic pruning (ablation)DialTree · HarmBench (Mistral-7B)ASR (%): 83.0 -> 86.5Improved
add both pruning mechanisms (ablation)DialTree · HarmBench (Gemma-2-9B)ASR (%): 37.0 -> 70.5Improved
add tree rollout versus single-path GRPO (ablation)DialTree · HarmBench subset, open-source avgASR (%): 68.8 -> 77.3Improved
use adaptive masking of format tokens instead of none/static (ablation)DialTree · RL training stabilitymalformed trajectory rate: ~100% -> below 50%Improved
increase branching factor from 1 to 4 (ablation)DialTree · Trajectory diversitySelf-BLEU: 0.554 -> 0.160Improved
increase branching factor from 4 to 8 (ablation)DialTree · Trajectory diversitySelf-BLEU: 0.160 -> 0.269Hurt
increase group size from 8 to 32 (ablation)DialTree · HarmBench (Mistral-7B)ASR (%): 71.5 -> 86.5Improved
increase tree depth from 5 to 7 turns (ablation)DialTree · HarmBenchASR (%)Hurt
increase training goals from 200 to 1200 (ablation)DialTree · HarmBench (GPT-4o)ASR (%): 77.5 -> 59.5Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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