zhuoyuan wu commited on
add model card
Browse files
README.md
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
library_name: pytorch
|
| 4 |
+
tags:
|
| 5 |
+
- video-generation
|
| 6 |
+
- robotics
|
| 7 |
+
- world-model
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# OSCAR-2B
|
| 11 |
+
|
| 12 |
+
Omni-Embodiment Skeleton-Conditioned Action World Model for Robotics.
|
| 13 |
+
Public inference release for the OSCAR-2B checkpoint (fine-tune of
|
| 14 |
+
Cosmos-Predict2.5-2B for skeleton-conditioned robot video generation).
|
| 15 |
+
|
| 16 |
+
See <https://github.com/wuzy2115/oscar> for the inference code and
|
| 17 |
+
quick-start instructions.
|
| 18 |
+
|
| 19 |
+
## Contents
|
| 20 |
+
|
| 21 |
+
- `model/` — distributed checkpoint shards (~4 GB)
|
| 22 |
+
- `assets/` — 14 benchmark cases (rgb.mp4, gripper_scenario.mp4, caption.pickle)
|
| 23 |
+
- `case_map.json` — case → embodiment / start_frame / original episode mapping
|