Alphabell.
Daily edition

Radar, 20 Nov 2026

Today's edition highlights the expanding scope of recursive self-improvement, from autonomous post-training pipelines to coding agents tuning optical hardware.

PaperarXiv·20 Nov 2026·Training data

AutoMaster: Autonomous Post-Training by Making the Same Model a Better Teacher

The loop: An LLM agent analyzes teacher trajectories to design skills that suppress undesirable habits and elicit effective behaviors. This improves the teacher model, which then generates higher quality data to fine-tune a better successor.

An autonomous post-training system uses an LLM agent to develop interventions that improve a teacher model's rollouts, yielding better fine-tuning data for target models.

loop fit 10/10Guoxin Chen, Shuang Sun, Fanzhe Meng, Xinyu Tang, Ruihua Song, Wayne Xin Zhaovia arXiv
AutoMaster: Autonomous Post-Training by Making the Same Model a Better Teacher
Blog postOpenAI·19 Nov 2026·Tools and harness

How OpenAI's Performance Factory Keeps ChatGPT Fast

The loop: Codex-powered workflows automatically find and fix performance issues within ChatGPT. This improves the system's execution speed, which makes the models themselves faster and more efficient at running future workflows.

OpenAI details a system of automated workflows that identify and resolve performance bottlenecks to maintain fast inference speeds for ChatGPT.

loop fit 9/10via OpenAI
How OpenAI's Performance Factory Keeps ChatGPT Fast
PaperarXiv·19 Nov 2026·Hardware and chips

In Situ Training of Optical Neural Networks by Coding Agents

The loop: A coding agent trains an optical neural network directly from camera measurements by forming a physical understanding of the optics. This tunes the hardware elements, making the resulting optical network better at classification and image encoding.

A coding agent successfully trains optical neural networks from measurements alone without requiring per-device calibration or gradients.

loop fit 9/10Ilker Oguz, Colin Christian, Christophe Moser, Demetri Psaltis, Gordon Wetzsteinvia arXiv
In Situ Training of Optical Neural Networks by Coding Agents
PaperarXiv·20 Nov 2026·Self-modifying agents

Designer-RSI: Evolving Procedural Memory for Graphic Design Agent

The loop: A graphic design agent evolves a modular skill bank from its own unlabeled design trajectories by contrasting successful and failed runs. This improves the agent's tool execution, making the same system better at fulfilling complex creative briefs.

A framework for recursive self-improvement in graphic design agents that evolves procedural memory from unlabeled trajectories across hundreds of design tools.

loop fit 9/10Hongyang Du, Lan Yan, Christian Flores, Asim Kadavvia arXiv
Designer-RSI: Evolving Procedural Memory for Graphic Design Agent
PaperarXiv·20 Nov 2026·Self-modifying agents

How Much Should an Agent Remember? Deployment Lessons from Self-Distilled Memory for LLM Agents

The loop: A pipeline mines lessons from an agent's own history and consolidates them into a smaller set of guidelines. This improves the agent's procedural memory, making the same system better at completing long-horizon tasks.

A deployed pipeline that allows an agent to learn from its own history by mining and consolidating lessons into reusable guidelines.

loop fit 9/10K. R. Jayaram, Vatche Isahagian, Aswath Vaithinathan Aravindan, Vinod...via arXiv
How Much Should an Agent Remember? Deployment Lessons from Self-Distilled Memory for LLM Agents
ProjectGitHub·18 Nov 2026·Kernels and compilers

baidu-baige/KODA

The loop: An open-weight agent uses compiler, correctness, and profiling feedback to optimize CUDA kernels. This improves the execution efficiency of the underlying hardware, which accelerates the agent's own future operations.

An open-source model and agent designed for CUDA kernel optimization using iterative feedback from compilers and profilers.

loop fit 8/10baidu-baigevia GitHub
baidu-baige/KODA
ProjectGitHub·18 Nov 2026·Self-modifying agents

chenkezhen480/evolution-computer-use

The loop: A computer-use agent learns and repairs its own procedures through interaction. This improves its library of executable skills, making the agent better at navigating desktop and browser environments.

An open-source self-evolving computer use agent that learns and repairs procedures for browser and Windows desktop navigation.

loop fit 7/10chenkezhen480via GitHub
chenkezhen480/evolution-computer-use

← 19 Nov 2026 Latest