KyutaiCADRE
CADRE action probes
CADRE action probes estimate actions from successive video frames. The four trained probes support evaluation of the robot, driving, indoor-navigation and human-manipulation world models.
Research release
The probes measure action consistency
A probe reads motion from generated frames so the researchers can compare it with the requested action. It is not a robot-control policy. The human-manipulation action-recovery measurement uses 12 wrist values rather than all 102 hand-pose inputs.cadre-action-probes: Kyutai model card
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Research release |
| Access | Downloadable |
| Model identifiers | kyutai/cadre-action-probes |
| License | CC BY-NC 4.0; DINOv3 terms also apply to the included frozen encoder.cadre-action-probes: Kyutai model card |
Related models
- CADRE 2x16x16 world models
Kyutai ยท CADRE
Four separately trained world models that predict video from context frames and actions in distinct environments.
Launch date not confirmedResearch releaseDownloadable