Do Machines Feel Lonely? Exploring the Ghost of Affection in Reinforcement Learning…

What happens when machine coordination starts to resemble human affection — and what it reveals about ethics, design, and ourselves 🧠 In a glowing data center humming with millions of…

REASONING GYM: Reasoning Environments for Reinforcement Learning with Verifiable Rewards

We introduce Reasoning Gym (RG), a library of reasoning environments for reinforcement learning with verifiable rewards. It provides over 100 data generators and verifiers spanning multiple domains including algebra, arithmetic, computation, cognition, geometry, graph theory, logic, and various common games. Its key innovation is the ability to generate virtually infinite training data with adjustable complexity, unlike most previous reasoning datasets, which are typically fixed. This procedural generation approach allows for continuous evaluation across varying difficulty levels. Our experimental results demonstrate the efficacy of RG in both evaluating and reinforcement learning of reasoning models.

Sutton & Barto Book: Reinforcement Learning: An Introduction

From PPO to GRPO - Hassan - Medium

Getting large language models (LLMs) to do what we really want isn’t always easy. Supervised learning helps, but it only gets you so far especially when you care about preferences, reasoning quality…

Reinforcement Learning Made Simple: Build a Q-Learning Agent in Python | Towards Data Science

Inspired by AlphaGo’s Move 37 — learn how agents explore, exploit, and win

Reinforcement Learning simplifica criação de agente Q-Learning

São Paulo — InkDesign News — No mundo da inteligência artificial, técnicas como machine learning e deep learning estão revolucionando a maneira como interagimos com a tecnologia. Um exemplo notável é

Outcome-based Reinforcement Learning to Predict the Future

Reinforcement learning with verifiable rewards (RLVR) has boosted math and coding in large language models, yet there has been little effort to extend RLVR into messier, real-world domains like forecasting. One sticking point is that outcome-based reinforcement learning for forecasting must learn from binary, delayed, and noisy rewards, a regime where standard fine-tuning is brittle. We show that outcome-only online RL on a 14B model can match frontier-scale accuracy and surpass it in calibration and hypothetical prediction market betting by adapting two leading algorithms, Group-Relative Policy Optimisation (GRPO) and ReMax, to the forecasting setting. Our adaptations remove per-question variance scaling in GRPO, apply baseline-subtracted advantages in ReMax, hydrate training with 100k temporally consistent synthetic questions, and introduce lightweight guard-rails that penalise gibberish, non-English responses and missing rationales, enabling a single stable pass over 110k events. Scaling ReMax to 110k questions and ensembling seven predictions yields a 14B model that matches frontier baseline o1 on accuracy on our holdout set (Brier = 0.193, p = 0.23) while beating it in calibration (ECE = 0.042, p < 0.001). A simple trading rule turns this calibration edge into \$127 of hypothetical profit versus \$92 for o1 (p = 0.037). This demonstrates that refined RLVR methods can convert small-scale LLMs into potentially economically valuable forecasting tools, with implications for scaling this to larger models.

Before you continue to YouTube