single mention · post-training
Single mixed-domain RL run
Trains coding, general-agent, visual, and cybersecurity tasks in one mixed RL run rather than separate domain runs.
The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See reinforcement learning algorithm for the methods filed beside it.
Filed under post-training › reinforcement learning algorithm