RSSAmplifier

Blog

Yingru Li

Yingru Li

richardli.xyzRSS feed ↗17 posts

Latest posts

Beyond Precision: Why Training-Inference Mismatch is an Optimization Problem and How Simple LR Scheduling Fixes It

Corresponding Author: Yingru Li Co-First Authors: Yaxiang Zhang and Yingru Li TL;DR The Problem: Reinforcement Learning (RL) training for LLMs is notoriously unstable. While recent studies attribute this to “training-inference mismatch” (caused by hybrid engines), standard fixes like Importance Sampling might fail during longer training runs. The Insight: We analyze this instability…

The Optimal Token Baseline

Project Lead: Yingru Li Co-First Authors: Yingru Li and Jiawei Xu TL;DR The Problem : RL training for LLMs frequently suffers from “training collapse” due to exploding gradient variance in long-horizon tasks. Standard baselines (like Group Mean) fail because they treat all tokens and sequences as equally “noisy.” The Insight : Gradient noise is heterogeneous. We derive the…

Trust Region Masking for Long-Horizon LLM Reinforcement Learning

Authors: Yingru Li, Jiacai Liu, Jiawei Xu, Yuxuan Tong, Ziniu Li, Baoxiang Wang 📄 arXiv: https://arxiv.org/abs/2512.23075 📊 Slides: trust_region_masking_slides_yingru.pdf 🔧 PR #4544: feat: trust region sequence masking Abstract Policy gradient methods for large language models optimize a surrogate objective computed from samples of a rollout policy $\pi_{\mathrm{roll}}$. When…

The Stability Gap: Why Top-K Routing Breaks RL Optimization

Citation @online { li-2025-topk-stability , title = {The Stability Gap: Why Top-K Routing Breaks {RL} Optimization} , author = {Li, Yingru} , year = {2025} , month = dec , url = {https://richardli.xyz/post/topk-routing-stability-gap/} } The Problem Training Mixture of Experts (MoE) language models with Reinforcement Learning can be unstable. While dense LLMs have continuous and differentiable…

Scalable Exploration via Ensemble++

The Exploration Challenge in Deep RL Thompson Sampling is one of the most elegant algorithms for balancing exploration and exploitation in sequential decision-making. The idea is simple: maintain a posterior distribution over reward functions, sample from it, and act optimally with respect to the sample. This approach naturally trades off exploring uncertain regions against exploiting known good…

Language as a Universal Interface for Reinforcement Learning Agents

Citation @article { li2025languageagent , title = {Language as a Universal Interface for Reinforcement Learning Agents} , author = {Li, Yingru} , journal = {Richard Li's Blog} , year = {2025} , url = {https://richardli.xyz/post/language-rl-agent/} } Introduction Why do some agents succeed while others fail? While frameworks like ReAct (Yao et al., 2023) show that interleaving reasoning and acting…

Mathematical Formulations of Rollout Correction Methods

Author: Yingru Li Abstract This document provides the definitive mathematical formulations for rollout correction methods in verl , following the natural progression from REINFORCE to PPO to Decoupled PPO . Rollout correction provides a unified framework to handle general off-policy problems in RL training - any scenario where the data collection distribution differs from the training…

Part 3: Trust Region Optimization via Sequence Masking

Authors: Yingru Li , Jiacai Liu Original Blog: When Speed Kills Stability: Demystifying RL Collapse from the Training-Inference Mismatch Series Context Part 1 : We established the SGA (Stochastic Gradient Ascent) framework and identified two failure modes of off-policy mismatch: Bias (measured by $D_{TV}$) and Variance (measured by $\chi^2$-divergence). Part 2 : We analyzed gradient estimators and…

Part 2: Applying the SGA Framework — Token v.s. Sequence-level Correction

Authors: Yingru Li , Jiacai Liu Original Blog: When Speed Kills Stability: Demystifying RL Collapse from the Training-Inference Mismatch Citation @online { liu-li-2025-rl-collapse , title = {When Speed Kills Stability: Demystifying {RL} Collapse from the Training-Inference Mismatch} , author = {Liu, Jiacai and Li, Yingru and Fu, Yuqian and Wang, Jiawei and Liu, Qian and Shen, Yu} , year = {2025} ,…

Part 1: Why Off-Policy Breaks RL — An SGA Analysis Framework

Authors: Yingru Li , Jiacai Liu Original Blog: When Speed Kills Stability: Demystifying RL Collapse from the Training-Inference Mismatch The Problem In reinforcement learning, we often cannot sample directly from the policy $\pi_\theta$ we are optimizing. Instead, we sample from a different behavior policy $\mu$. This off-policy setting ($\mu \neq \pi$) arises from multiple sources: Standard…

Information Bandwidth in Reinforcement Learning

How Gradient Structure Determines Information Capacity When I first read the “ LoRA Without Regret ” blog post, one claim caught my attention: policy gradient algorithms learn roughly 1 bit of information per episode . This insight elegantly explains why LoRA—with its mere thousands of trainable parameters—works so remarkably well for RL fine-tuning of large language models. But what…

When Speed Kills Stability: Demystifying RL Collapse from the Training-Inference Mismatch

Co-First Authors: Jiacai Liu and Yingru Li Corresponding Authors: Yingru Li and Yu Shen TL;DR The relentless push for faster inference has created a dangerous “training-inference mismatch” that can silently kill reinforcement learning with LLMs. Our investigation reveals a vicious cycle that is particularly acute in modern reasoning and agentic RL: OOD Contexts Drive Low-Probability…

HyperAgent - A Simple, Efficient, Scalable and Provable RL Framework

HyperAgent - A Simple, Efficient and Scalable RL Framework for Complex Environments

Towards AGI for Humanity through Efficient Reinforcement Learning

No-Regret Learning in Unknown Game with Applications

HyperDQN - Randomized Exploration for Deep Reinforcement Learning