autoresearch
6 mentions across all digests
Autoresearch is an agentic framework pioneered by Andrej Karpathy where an AI agent autonomously runs iterative ML experiments — editing code, training, evaluating, and committing or reverting — demonstrated to improve LLM training efficiency by ~11% across ~700 experiments.
[AINews] Autoresearch: Sparks of Recursive Self Improvement
Collaborative Autoresearch on a Peer-to-Peer Network
Autonomous research systems can operate collaboratively across peer-to-peer networks, eliminating dependence on centralized infrastructure for distributed knowledge work.
MAGNET: Autonomous Expert Model Generation via Decentralized Autoresearch and BitNet Training
MAGNET automates creation of task-specific expert AI models through decentralized autoresearch and BitNet quantization, enabling efficient autonomous agent development without manual specialization.
Autoresearch on an old research idea
Claude Code autonomously optimized eCLIP genomics models through iterative training loops and architecture experiments, progressing from hyperparameter tuning to AI-generated novel research hypotheses.
Shopify/liquid: Performance: 53% faster parse+render, 61% fewer allocations
Shopify's CEO used agentic autoresearch patterns to run ~120 semi-autonomous experiments on Liquid, achieving 53% faster parse+render and 61% fewer allocations—demonstrating how agents unlock high-ROI optimization work that's impractical for humans to tackle manually.