Interpreting and Controlling Model Behavior via Constitutions for Atomic Concept Edits
cs.LG, cs.AI, cs.CL, cs.CV
Submitted: 2026-01-23
Updated: 2026-11-16
Terminology
Sources
- Generating Natural Language Adversarial Examples
- Constitutional AI: Harmlessness from AI Feedback
- Explaining Image Classifiers by Counterfactual Generation
- Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities
- Towards A Rigorous Science of Interpretable Machine Learning
- Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context
- Explaining Classifiers with Causal Concept Effect (CaCE)
- Active Evaluation Acquisition for Efficient LLM Benchmarking
- Using Captum to Explain Generative Language Models
- DecompX: Explaining Transformers Decisions by Propagating Token Decomposition
- AlphaEvolve: A coding agent for scientific and algorithmic discovery
- GPT-4 Technical Report
- Red Teaming Language Models with Language Models
- Automatic Prompt Optimization with "Gradient Descent" and Beam Search
- Zero-Shot Text-to-Image Generation
- Position: Adversarial ML for LLMs Is Not Making Any Progress
- Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding
- Intriguing properties of neural networks
- TextGrad: Automatic "Differentiation" via Text
- Universal and Transferable Adversarial Attacks on Aligned Language Models
Related papers
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis
- (Mis)generalization of Helpful-Only Fine-tuning