Previews re-rendered at 1024x1024 from the recorded demonstrations

Dataset card: 85% overlap criterion for every task

Dataset card: per-task success criteria

Dataset card: separate gif preview table

Dataset card: task gifs, paper citation, drop asym

Remove the asym extra task

Dataset card for the CAMP release (task names follow robo-camp.github.io)

Rename task folders to the paper's task names (Multi-Goals, Swap-Direct, Swap-Shuffle, Find-Track, Asym)

Update

README: lowercase first letter and drop trailing period in Memory requirement column

Add per-step data/language_instruction (constant per task); add Description column to README

README: fold per-task memory requirements into the Tasks table; remove duplicate per-task sections

Truncate data/state to first 2 dims across all tasks; uniform observation schema

Initial release: 5 tasks, observations only (no privileged hidden-state arrays)

initial commit