CodeMidas creates RL training tasks from code
In a post dated September 21, 2026, Xiaomi describes CodeMidas, which uses agents to generate 5,545 executable tasks in 23 languages from open-source codebases.
Source: CodeMidas gera tarefas de treinamento de RL a partir de código (x.com). Text prepared with AI from this source.
What happened and what to do
In a post dated September 21, 2026, Xiaomi presents CodeMidas, which uses agents to extract executable tasks from open-source codebases without relying on issues or commits. The approach generated 5,545 tasks across 23 languages and can create reinforcement learning (RL) training environments directly from source code.
A company could adapt this approach to turn its repositories into evaluation and training tasks, with execution criteria, tests, and quality monitoring. A project could begin by selecting codebases, defining access controls, and validating the safety and usefulness of generated tasks before incorporating them into development or training processes.
How the consultancy can help
Wendelmaques can assess the opportunity and risks in a company's repositories, define the scope, and implement pipelines for task collection, generation, and validation, as well as monitoring and operation.
Next step
Send a short description of your repositories, goals, and constraints to receive a scoped proposal.
Let's assess your project
I reply with scope and a proposal.