An open-source agent framework says it automates several stages of reinforcement learning research workflows, including GPU experiments.
Published on June 17, 2026, Deli AutoResearch is described as an open-source agent framework for reinforcement learning (RL) research workflows. The author says it autonomously planned GPU experiments and automated design, code writing, execution, debugging, and summarizing conclusions in RL runs described for a 285B model.
The project provides an example of an agent carrying out several stages of an RL research workflow. To assess what was demonstrated, consult the original publication and inspect its example and reported results; the available description does not specify metrics or experimental conditions.