CosmoCore Affective Dream-Replay Reinforcement Learning for Code Generation

TL;DR

CosmoCore is a neuroscience-inspired reinforcement learning architecture that uses affective signals to improve code generation by prioritizing error correction, reducing hallucinations, and accelerating learning in large language models.

Contribution

It introduces a novel affective RL framework that incorporates valence and surprise signals to enhance code generation, extending beyond traditional human feedback methods.

Findings

Reduces hallucinated code by 48%

Accelerates self-correction by 45%

Validates effectiveness on multiple benchmarks

Abstract

We introduce CosmoCore, a neuroscience-inspired reinforcement learning (RL) architecture that integrates affective signals to enhance code generation in large language models (LLMs). Motivated by human and animal learning where embarrassment from mistakes drives rapid correction, as observed in training a puppy to avoid repeating errors after a single scolding CosmoCore tags code generation trajectories with valence and surprise using a lightweight multi-layer perceptron (MLP). High-negative valence (cringe) episodes, such as buggy code outputs, are prioritized in a Dream Queue for five-fold replay during off-policy updates, while low-surprise successes are pruned to prevent overconfidence and buffer bloat. Evaluated on code generation benchmarks like HumanEval and BigCodeBench, alongside simulations with a custom data pipeline environment, CosmoCore reduces hallucinated code (e.g., syntax errors or logical bugs) by 48\% and accelerates self-correction by 45\%. Local experiments using Hugging Face models in a PySpark environment validate these gains, with code snippets provided for replication. Ablations confirm valence tagging boosts curiosity in exploration, and pruning mitigates inefficiency. This framework extends RL from human feedback (RLHF) for more emotionally aware code assistants, with applications in IDEs and data pipelines. Code and the custom mini-world simulation are released.