Does OpenVLA run on ALOHA (stationary)?
Scope: Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.
Capabilities
| Capability | State | Notes | Evidence |
|---|---|---|---|
| Train / fine-tune | works | ev-hub-anjali-openvla-lora, ev-hub-anjali-openvla-full | |
| Real-robot inference | no verified evidence | — | |
| Simulation inference | no verified evidence | — | |
| Teleop + record | no verified evidence | — |
Evidence 2 sources; the cell shows the highest tier
- T3 · community reportedhub-checkpointsource 2026-08-01verified 2026-10-25 (today)
Model card: LoRA (r=32) on openvla/openvla-7b fine-tuned on 48 episodes of lerobot/aloha_static_coffee (meta/info.json robot_type=aloha, real static ALOHA, 50 episodes), using only cam_left_wrist and 14-D joint targets discretised into 256 bins. Card states it is not validated on a robot or in simulation and fails both held-out episodes offline. Not a LeRobot-format checkpoint (no config.json policy type).
https://huggingface.co/anjali209/openvla-7b-aloha-coffee-lora - T3 · community reportedhub-checkpointsource 2026-08-01verified 2026-10-25 (today)
Model card: full fine-tune of openvla/openvla-7b on the same 48 lerobot/aloha_static_coffee episodes (single wrist camera); offline teacher-forced evaluation only, same stated limitations.
https://huggingface.co/anjali209/openvla-7b-aloha-coffee-full
Embed this cell
Badge for READMEs and model cards. It always shows the current state.
[](https://caniembody.pages.dev/compat/aloha/openvla)<iframe src="https://caniembody.pages.dev/embed/aloha/openvla" width="380" height="132" style="border:0" title="OpenVLA on ALOHA (stationary) | CanIEmbody"></iframe>History
No committed history yet. Source file: data/compat/aloha__openvla.yaml