Add files using upload-large-folder tool
Browse files- README.md +36 -3
- eva_i2v_14B.ckpt +3 -0
README.md
CHANGED
|
@@ -1,3 +1,36 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
library_name: none
|
| 4 |
+
pipeline_tag: image-to-video
|
| 5 |
+
tags:
|
| 6 |
+
- image-to-video
|
| 7 |
+
- robotics
|
| 8 |
+
- world-model
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# EVA
|
| 12 |
+
|
| 13 |
+
This repository hosts the EVA checkpoint released with:
|
| 14 |
+
|
| 15 |
+
**EVA: Aligning Video World Models with Executable Robot Actions via Inverse Dynamics Rewards**
|
| 16 |
+
|
| 17 |
+
Project page: https://eva-project-page.github.io/
|
| 18 |
+
|
| 19 |
+
## Model Summary
|
| 20 |
+
|
| 21 |
+
This checkpoint is an EVA model for robotic video planning.
|
| 22 |
+
|
| 23 |
+
It is built on top of a Wan2.1 I2V 14B backbone and further adapted on **Robotwin** through:
|
| 24 |
+
|
| 25 |
+
- supervised fine-tuning (SFT)
|
| 26 |
+
- reinforcement-learning-based post-training (RL)
|
| 27 |
+
|
| 28 |
+
The released checkpoint corresponds to the merged model after both stages of post-training.
|
| 29 |
+
|
| 30 |
+
## Intended Use
|
| 31 |
+
|
| 32 |
+
This model is intended for research use in robot video prediction and visual planning.
|
| 33 |
+
|
| 34 |
+
Given an input image and a language instruction, the model generates future video rollouts that are better aligned with executable robot behavior.
|
| 35 |
+
|
| 36 |
+
|
eva_i2v_14B.ckpt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d9deaf46ada7ac0d0ddb311ce7ce9f480ab74dfadfee58bb8c983ebbf5e83f25
|
| 3 |
+
size 32790200913
|