📄️ Introduction
What is the ADK?
📄️ Agents
Explanation
📄️ Models
Explanation
📄️ Model handling
Models store per-system checkpoints and metadata under the models directory of an agent. During an optimization run, your agent loads weights from /models/ via load_models. Model handling covers everything beyond that day-to-day loop: cloning a model for a new circuit, exporting artifacts off the platform, and importing them back.
📄️ Specifications
Explanation
📄️ Environments
Gymnasium environments in the ADK are part of the RL path when you use
📄️ RL Agents
An RL agent in the ADK is your RLAgentEnv subclass — the class
📄️ RL Run Data
When you use RLExecutor, it builds two data objects from the optimization
📄️ Optimization loop
Every optimization run eventually loops until stopped, a limit is reached, or (in inference) targets
📄️ Agent Settings
Agent metadata and optional runtime options live in settings.json at the project root. The file is
📄️ Environment Settings
Local runtime configuration for the ADK lives in .env at the project root. This file is created by