7 papers
How Neural Reward Models Learn Features for Policy Optimization: A Single-Index Analysis
Rei Higuchi, Ryotaro Kawata, Akifumi Wachi +3
Reward modeling is not only a prediction problem: in KL-regularized policy optimization, the learned reward is exponentiated to define the deployed policy, so downstream value depe…
Inference-Aware Meta-Alignment of LLMs via Non-Linear GRPO
Shokichi Takakura, Akifumi Wachi, Rei Higuchi +2
Aligning large language models (LLMs) to diverse human preferences is fundamentally challenging since criteria can often conflict with each other. Inference-time alignment methods…
A Relative-Budget Theory for Reinforcement Learning with Verifiable Rewards in Large Language Model Reasoning
Akifumi Wachi, Hirota Kinoshita, Shokichi Takakura +2
Reinforcement learning (RL) is a dominant paradigm for improving the reasoning abilities of large language models, yet its effectiveness varies across tasks and compute budgets. We…
When Does Metadata Conditioning (NOT) Work for Language Model Pre-Training? A Study with Context-Free Grammars
Rei Higuchi, Ryotaro Kawata, Naoki Nishikawa +7
The ability to acquire latent semantics is one of the key properties that determines the performance of language models. One convenient approach to invoke this ability is to prepen…
Degrees of Freedom for Linear Attention: Distilling Softmax Attention with Optimal Feature Efficiency
Naoki Nishikawa, Rei Higuchi, Taiji Suzuki
Linear attention has attracted interest as a computationally efficient approximation to softmax attention, especially for long sequences. Recent studies have explored distilling so…
Direct Density Ratio Optimization: A Statistically Consistent Approach to Aligning Large Language Models
Rei Higuchi, Taiji Suzuki
Aligning large language models (LLMs) with human preferences is crucial for safe deployment, yet existing methods assume specific preference models like Bradley-Terry model. This a…