No article text was recorded for this version.
cross-scale model steering
Version 2 of 2102 papers133 conceptsfrozen 2026-10-04
Papers in this version
- Blog: Weak-to-Strong Search: Align Large Language Models via Searching over Small Language Models
- Blog: DIESEL: A Lightweight Inference-Time Safety Enhancement for Language Models
- Blog: REFINER: Reasoning Feedback on Intermediate Representations
- Blog: Using RL to Identify Divisive Perspectives Improves LLMs Abilities to Identify Communities on Social Media
- Blog: Contextualized Privacy Defense for LLM Agents
- Blog: Towards Resource Efficient and Interpretable Bias Mitigation in Large Language Models
- Blog: GenARM: Reward Guided Generation with Autoregressive Reward Model for Test-time Alignment
- Blog: $Q\sharp$: Provably Optimal Distributional RL for LLM Post-Training
- Blog: Personalized Abstractive Summarization by Tri-agent Generation Pipeline
- Blog: DExperts: Decoding-Time Controlled Text Generation with Experts and Anti-Experts
- Blog: Activation Space Interventions Can Be Transferred Between Large Language Models
- Blog: Dynamic Evil Score-Guided Decoding: An Efficient Decoding Framework For Red-Team Model
- Blog: Systematic Rectification of Language Models via Dead-end Analysis
- Blog: Overcoming Multi-step Complexity in Multimodal Theory-of-Mind Reasoning: A Scalable Bayesian Planner
- Blog: Nudging: Inference-time Alignment of LLMs via Guided Decoding
- Blog: Aligning Frozen LLMs by Reinforcement Learning: An Iterative Reweight-then-Optimize Approach
- Blog: RA-ISF: Learning to Answer and Understand from Retrieval Augmentation via Iterative Self-Feedback
- Blog: RL4F: Generating Natural Language Feedback with Reinforcement Learning for Repairing Model Outputs
- Blog: Instance Needs More Care: Rewriting Prompts for Instances Yields Better Zero-Shot Performance
- Blog: Inference-Time Language Model Alignment via Integrated Value Guidance
- Blog: Overcoming Multi-step Complexity in Multimodal Theory-of-Mind Reasoning: A Scalable Bayesian Planner
- Blog: PARM: Multi-Objective Test-Time Alignment via Preference-Aware Autoregressive Reward Model
- Blog: On-the-Fly Controlled Text Generation with Experts and Anti-Experts
- Blog: DeepCritic: Deliberate Critique with Large Language Models
- Blog: PARM: Multi-Objective Test-Time Alignment via Preference-Aware Autoregressive Reward Model
- Blog: Reward-Shifted Speculative Sampling Is An Efficient Test-Time Weak-to-Strong Aligner
- Blog: Alleviating Hallucinations of Large Language Models through Induced Hallucinations
- Blog: Cross-model Transferability among Large Language Models on the Platonic Representations of Concepts
- Blog: UCD: Unlearning in LLMs via Contrastive Decoding
- Blog: Tuning Language Models by Proxy
- Blog: Truthfulness Despite Weak Supervision: Evaluating and Training LLMs Using Peer Prediction
- Blog: Instruction Induction: From Few Examples to Natural Language Task Descriptions
- Blog: Weak-to-Strong Jailbreaking on Large Language Models
- Blog: MT-RewardTree: A Comprehensive Framework for Advancing LLM-Based Machine Translation via Reward Modeling
- Blog: Your Weak LLM is Secretly a Strong Teacher for Alignment
- Blog: Aligner: Efficient Alignment by Learning to Correct
- Blog: GenARM: Reward Guided Generation with Autoregressive Reward Model for Test-Time Alignment
- Blog: Scaling Autonomous Agents via Automatic Reward Modeling And Planning
- Blog: Proposer-Agent-Evaluator (PAE): Autonomous Skill Discovery For Foundation Model Internet Agents
- Blog: Teaching Language Models to Critique via Reinforcement Learning
- Blog: Agentic Monte Carlo: Simulating Reinforcement Learning for Black-Box Agents
- Blog: Advancing Zero-shot Text-to-Speech Intelligibility across Diverse Domains via Preference Alignment
- Blog: Stream Aligner: Efficient Sentence-Level Alignment via Distribution Induction
- Blog: RL4F: Generating Natural Language Feedback with Reinforcement Learning for Repairing Model Outputs
- Blog: Offset Unlearning for Large Language Models
- Blog: Reward-Augmented Decoding: Efficient Controlled Text Generation With a Unidirectional Reward Model
- Blog: ARGS: Alignment as Reward-Guided Search
- Blog: ProRefine: Inference-time Prompt Refinement with Textual Feedback
- Blog: A General Framework for Inference-time Scaling and Steering of Diffusion Models
- Blog: Leading the Follower: Learning Persuasive Agents in Social Deduction Games
- Blog: Post Hoc Explanations of Language Models Can Improve Language Models
- Blog: Self-Supervised Alignment with Mutual Information: Learning to Follow Principles without Preference Labels
- Blog: Token-Level Inference-Time Alignment for Vision-Language Models
- Blog: Mitigating Hallucination in Abstractive Summarization with Domain-Conditional Mutual Information
- Blog: GenSco: Can Question Decomposition based Passage Alignment improve Question Answering?
- Blog: DIESEL - Dynamic Inference-Guidance via Evasion of Semantic Embeddings in LLMs
- Blog: Activation Space Interventions Can Be Transferred Between Large Language Models
- Blog: Train-Attention: Meta-Learning Where to Focus in Continual Knowledge Learning
- Blog: Continuously Steering LLMs Sensitivity to Contextual Knowledge with Proxy Models
- Blog: Embedding-Aligned Language Models
Showing 60 of 102 papers.
History & citation
Current · Permanent link
Cite this direction
Latest version (a living page):
@misc{lacuna_cross_scale_model_steering_3ee5a12a,
title = {cross-scale model steering},
author = {{Lacuna}},
howpublished = {Lacuna research direction},
year = {2026},
url = {https://lacuna.tiptreesystems.com/direction/cross-scale-model-steering/rdr_3ee5a12a2f0445d3a177f3db5e30c321},
note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-22},
}This exact version (version 2, frozen text):
@misc{lacuna_cross_scale_model_steering_3ee5a12a_v2,
title = {cross-scale model steering},
author = {{Lacuna}},
howpublished = {Lacuna research direction},
year = {2026},
url = {https://lacuna.tiptreesystems.com/direction/cross-scale-model-steering/rdr_3ee5a12a2f0445d3a177f3db5e30c321/v/2},
note = {Version 2, frozen 2026-10-04. Accessed 2026-11-22},
}