Skip to main content
Config optimization involves systematically refining the parameters of various nodes within your agent. Specifically, this includes prompts, model choices, and other hyperparameters. To access config optimization settings, click on the Optimize button at the top of the Agent Composer canvas. It should open the following modal: Modal for config optimization settings Select the environment(s) you want to run the optimization in along with the number of rollouts per envionment. Then, set the hyperparameters for optimization:
​
Batch Size
integer
default:"8"
required
Base batch size to use for individual optimization steps
​
Explore Radius
integer
default:"5"
required
A positive integer controlling the aggressiveness of exploration during optimization. A larger explore radius encourages the optimizer to make more substantial changes between successive configurations.
​
Explore factor
float
default:"0.5"
required
A float between 0 to 1 controlling the exploration-exploitation trade-off. A higher explore factor allocates more rollouts to discover new configs, while a lower value allocates more rollouts to ensure the discovered configs are thoroughly evaluated.
Select the RELAI API key to use for reporting optimization metrics. Finally, select the parameters to be optimized. You can enable optimization for any parameter by toggling the Optimize switch next to it. You can also configure the search space for each parameter by clicking the gear icon next to it. This will open a modal like the following: Modal for configuring parameter search space In this modal, optionally, you can constrain the search space to a specific range or set of options depending on the parameter type. After configuring the optimization settings and parameters, click on Run Optimization to begin the process. You can monitor the status of the optimization on the Agent Jobs page. When the optimization is complete, you can view the results on the Maestro Runs page.