Static reports for RLBench closed-loop inference experiments.
- Report: docs/rlbench-comparison-dashboard/index.html
- Models: DP, DP3, LingBot RGB, Rynn, LingBot RGB-Depth-Flow-Action
- Controls: model buttons, task dropdown, step dropdown
- Missing models/runs are kept visible as empty states so new results can be added without changing the UI.
- Report: docs/dp-inference-report/index.html
- Baseline: Diffusion Policy
- Action mode: absolute
- Episodes: 10 seeds per task, seed base
930000
| Task | Success |
|---|---|
place_cups |
0/10 |
place_shape_in_shape_sorter |
0/10 |
stack_blocks |
0/10 |
stack_cups |
0/10 |
The report includes the per-task sweep_summary.json, metrics JSON, and all saved observation videos.
- Report: docs/lingbot-va-inference-report/index.html
- Variants:
rgb-action,rgb-depth-flow-action - Episodes: 10 seeds per completed run, seed base
930000 - Current completed latest-run total:
0/50
The report includes completed and in-progress run status, per-episode outcome tables, and selected final comparison videos.