ARL-RR alternates optimization over rubric meta-classes with dynamic selection to avoid fixed scalarization, outperforming baselines on HealthBench.
Gradient-adaptive policy optimization: Towards multi-objective alignment of large language models
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
representative citing papers
VC-Soup uses a cosine-similarity consistency metric to filter data, trains value-consistent policies, and applies linear merging with Pareto filtering to improve multi-value LLM alignment trade-offs.
A survey compiling RL methods, challenges, data resources, and applications for enhancing reasoning in large language models and large reasoning models since DeepSeek-R1.
citing papers explorer
-
Alternating Reinforcement Learning with Contextual Rubric Rewards: Beyond the Scalarization Strategy
ARL-RR alternates optimization over rubric meta-classes with dynamic selection to avoid fixed scalarization, outperforming baselines on HealthBench.
-
VC-Soup: Value-Consistency Guided Multi-Value Alignment for Large Language Models
VC-Soup uses a cosine-similarity consistency metric to filter data, trains value-consistent policies, and applies linear merging with Pareto filtering to improve multi-value LLM alignment trade-offs.
-
A Survey of Reinforcement Learning for Large Reasoning Models
A survey compiling RL methods, challenges, data resources, and applications for enhancing reasoning in large language models and large reasoning models since DeepSeek-R1.