Context-Grounding Gains Are Mediated by Pre-existing Machinery: Auditing GRPO, SFT, and DPO
cs.CL, cs.AI, cs.LG
Submitted: 2026-09-01
Updated: 2026-09-01
Terminology
Sources
- Better World Models Can Lead to Better Post-Training Performance
- How Does Alignment Tuning Shape Representations of Sycophancy and Related Cue-Induced Biases in LLMs?
- ContextFocus: Activation Steering for Contextual Faithfulness in Large Language Models
- SHIFT: Gate-Modulated Activation Steering for Knowledge Conflict Mitigation in Retrieval-Augmented Generation
- Evaluating GRPO and DPO for Faithful Chain-of-Thought Reasoning in LLMs
- Vision-Default, Prior-Override: Causal Mechanisms of Perception-Knowledge Conflict in Vision-Language Models
- Emergent Misalignment Recruits a Pre-existing Persona Subspace
- The Hydra Effect: Emergent Self-repair in Language Model Computations
- Towards Reliable Evaluation of Behavior Steering Interventions in LLMs
- Actionable Activation Directions for Detecting and Mitigating Emergent Misalignment Across Language Model Families
- EvidenceRL: Reinforcing Evidence Consistency for Trustworthy Language Models
- Steering Language Models With Activation Engineering
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- Emergence of Context Characteristics Sensitivity in Large Language Models
- Representation Engineering: A Top-Down Approach to AI Transparency
- Reasoning-Finetuning Repurposes Latent Representations in Base Models
- The Invisible Leash: Why RLVR May or May Not Escape Its Origin
- Demystifying Variance in Circuit Discovery of LLMs
Related papers
- Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
- Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
- Subliminal Steering: Stronger Encoding of Hidden Signals
- MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
- The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
- Untangling the Mechanisms of Misleading Context in Medical Question Answering