mc-training-capture

Automating Minecraft training data capture involves enabling a script that records partial observation grids and agent context at configured movement intervals. This Skill toggles that automatic recording process to build datasets for perception models.

Yes, ground truth data can be included when capturing Minecraft samples by integrating a specific bridge URL. The Skill fetches detailed environmental ground truth from this endpoint to create more comprehensive training sets.

Sample intervals for recording Minecraft world states are configured based on agent movement. You can set the specific interval distance, triggering automatic data capture and observation grid recording only when the agent moves.

Fetching ground truth data during Minecraft capture requires a Minecraft bridge URL. The Skill integrates with this bridge endpoint to automatically pull and include detailed ground truth information alongside your captured observation grids.

Yes, this Minecraft data capture method works without external dependencies. It operates using internal scripts to manage the recording of observation grids and bridge URL integration, requiring no additional packages to function.

Automated capture ensures consistent sampling intervals and immediate ground truth integration for Minecraft training data. It removes manual effort, reliably recording partial observation grids precisely when the agent moves to streamline perception model training.