First-Order Steering: Translating Weight Adaptation into Activation Steering
cs.LG, cs.AI, cs.CL
Submitted: 2026-10-03
Updated: 2026-10-03
Terminology
Sources
- Gemma 4 Technical Report
- LANCE: Low Rank Activation Compression for Efficient On-Device Continual Learning
- Qwen2.5-VL Technical Report
- Fluctuations of eigenvalues and second order Poincar'e inequalities
- Enhancing Chat Language Models by Scaling High-quality Instructional Conversations
- LoRA: Low-Rank Adaptation of Large Language Models
- Mistral 7B
- Adam: A Method for Stochastic Optimization
- LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods
- Deployable Per-Instance Multi-Layer Activation Steering for Large Language Models
- LoRA Soups: Merging LoRAs for Practical Skill Composition Tasks
- Controlling Language and Diffusion Models by Transporting Activations
- A Survey on Sparse Autoencoders: Interpreting the Internal Mechanisms of Large Language Models
- Model merging with SVD to tie the Knots
- Steering Language Models With Activation Engineering
- Extending Activation Steering to Broad Skills and Multiple Behaviours
- Representation Engineering: A Top-Down Approach to AI Transparency
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks