Update README.md
Browse files
README.md
CHANGED
|
@@ -12,7 +12,11 @@ tags:
|
|
| 12 |
|
| 13 |
# Model Card for act
|
| 14 |
|
| 15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
|
| 17 |
|
| 18 |
[Action Chunking with Transformers (ACT)](https://huggingface.co/papers/2304.13705) is an imitation-learning method that predicts short action chunks instead of single steps. It learns from teleoperated data and often achieves high success rates.
|
|
|
|
| 12 |
|
| 13 |
# Model Card for act
|
| 14 |
|
| 15 |
+
ACT trained on PushT dataset for 80,000 training steps.
|
| 16 |
+
Trained for 3.5 hrs on T4 GPU on Colab free.
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
The goal is to use this to compare ACTs performance on PushT against diffusion policy particularly the aspect of action multimodality.
|
| 20 |
|
| 21 |
|
| 22 |
[Action Chunking with Transformers (ACT)](https://huggingface.co/papers/2304.13705) is an imitation-learning method that predicts short action chunks instead of single steps. It learns from teleoperated data and often achieves high success rates.
|