TRACESTEER: Trace Driven Activation Steering for Hallucination Mitigation in Diffusion Language Models
We propose TRACESTEER, a trace-driven activation steering framework that uses preceding denoising predictions as feedback to jointly adapt steering direction and strength.
ProofPaper ↗
Key points
- Diffusion large language models (dLLMs) have emerged as a promising alternative to autoregressive LLMs, yet remain susceptible to hallucination.
- Activation steering offers a promising mitigation approach by guiding parallel token predictions through internal representations.
- Trajectory-Guided Steering Gate (TS-Gate) regulates their strength using prediction uncertainty and the denoising timestep.
- With the backbone frozen, factual supervision and magnitude regularization jointly guide the learning of adaptive steering interventions, promoting factual reliability and stability throughout denoising.
Sources (1)
- [1]TRACESTEER: Trace Driven Activation Steering for Hallucination Mitigation in Diffusion Language ModelsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 13, 11:18 AM
We propose TRACESTEER, a trace-driven activation steering framework that uses preceding denoising predictions as feedback to jointly adapt steering direction and strength.
Diffusion large language models (dLLMs) have emerged as a promising alternative to autoregressive LLMs, yet remain susceptible to hallucination.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 13, 2026Mitigating Cross-Modal Hallucinations by Correcting Modality Misrouting in Omnimodal LLMs
- Oct 13, 2026Beyond the Verdict: Agentic Code Judging with Verifiable Feedback
- Oct 12, 2026Strong yet Controllable LLM Abstention without Learning to Abstain
- Oct 12, 2026AI Representations of Mental Health Stigma: Quantifying the Gap Between AI Estimations and Human Perceptions Across Gender Groups Using GPT-Based LLMs
- Oct 12, 2026Spectral-Driven Adaptive-Rank Optimizer for Large Language Model Reinforcement Learning with Verifiable Rewards
- Oct 12, 2026Large Language Models for Machine Translation Quality Annotation: Humans and Models Are Both Challenged
