Update README.md
Browse files
README.md
CHANGED
|
@@ -9,3 +9,27 @@ This repository contains the latest checkpoint of the training visible at: https
|
|
| 9 |
|
| 10 |
I am researching for more efficient ways of training diffusion and therefore I am experimenting with the architecture. As a result to replicate or use the model use this branch of "huggingface/lerobot": https://github.com/the-future-dev/lerobot/tree/cloth-diff
|
| 11 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
|
| 10 |
I am researching for more efficient ways of training diffusion and therefore I am experimenting with the architecture. As a result to replicate or use the model use this branch of "huggingface/lerobot": https://github.com/the-future-dev/lerobot/tree/cloth-diff
|
| 11 |
|
| 12 |
+
|
| 13 |
+
## Demo Video
|
| 14 |
+
Here’s a sample output from the model:
|
| 15 |
+
|
| 16 |
+
<video controls width="550">
|
| 17 |
+
<source src="https://huggingface.co/the-future-dev/diffusion-pusht-keypoints/resolve/main/replay.mp4" type="video/mp4">
|
| 18 |
+
Your browser does not support the video tag.
|
| 19 |
+
</video>
|
| 20 |
+
|
| 21 |
+
## Evaluation
|
| 22 |
+
|
| 23 |
+
The model was evaluated on the `PushT` environment from [gym-pusht](https://github.com/huggingface/gym-pusht). There are two evaluation metrics on a per-episode basis:
|
| 24 |
+
|
| 25 |
+
- Maximum overlap with target (seen as `eval/avg_max_reward` in the charts above). This ranges in [0, 1].
|
| 26 |
+
- Success: whether or not the maximum overlap is at least 95%.
|
| 27 |
+
|
| 28 |
+
Here are the metrics for 500 episodes worth of evaluation.
|
| 29 |
+
|
| 30 |
+
Metric|Average over 500 episodes
|
| 31 |
+
-|-
|
| 32 |
+
Average max. overlap ratio | 0.9780
|
| 33 |
+
Success rate (%) | 86.80%
|
| 34 |
+
|
| 35 |
+
The results of each of the individual rollouts may be found in [eval_results.json](eval_results.json).
|