Robotics
Transformers
Safetensors
Inference Endpoints
JayLee131 commited on
Commit
e2a4882
·
verified ·
1 Parent(s): ea89838

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +5 -3
README.md CHANGED
@@ -8,14 +8,14 @@ pipeline_tag: robotics
8
 
9
  VQ-BeT (as per [Behavior Generation with Latent Actions](https://arxiv.org/abs/2403.03181)) trained for the `PushT` environment from [gym-pusht](https://github.com/huggingface/gym-pusht).
10
 
11
- ![demo](demo.gif)
12
-
13
  ## How to Get Started with the Model
14
 
15
  See the [LeRobot library](https://github.com/huggingface/lerobot) (particularly the [evaluation script](https://github.com/huggingface/lerobot/blob/main/lerobot/scripts/eval.py)) for instructions on how to load and evaluate this model.
16
 
17
  ## Training Details
18
 
 
 
19
  The model was trained using this command:
20
 
21
  ```bash
@@ -40,7 +40,9 @@ The model was evaluated on the `PushT` environment from [gym-pusht](https://gith
40
  - Maximum overlap with target (seen as `eval/avg_max_reward` in the charts above). This ranges in [0, 1].
41
  - Success: whether or not the maximum overlap is at least 95%.
42
 
 
 
43
  <blank>|Ours
44
  -|-
45
- Average max. overlap ratio | 0.887
46
  Success rate for 500 episodes (%) | 66.0
 
8
 
9
  VQ-BeT (as per [Behavior Generation with Latent Actions](https://arxiv.org/abs/2403.03181)) trained for the `PushT` environment from [gym-pusht](https://github.com/huggingface/gym-pusht).
10
 
 
 
11
  ## How to Get Started with the Model
12
 
13
  See the [LeRobot library](https://github.com/huggingface/lerobot) (particularly the [evaluation script](https://github.com/huggingface/lerobot/blob/main/lerobot/scripts/eval.py)) for instructions on how to load and evaluate this model.
14
 
15
  ## Training Details
16
 
17
+ Trained with [LeRobot@342f429](https://github.com/huggingface/lerobot/tree/342f429f1c321a2b4501c3007b1dacba7244b469).
18
+
19
  The model was trained using this command:
20
 
21
  ```bash
 
40
  - Maximum overlap with target (seen as `eval/avg_max_reward` in the charts above). This ranges in [0, 1].
41
  - Success: whether or not the maximum overlap is at least 95%.
42
 
43
+ The results of each of the individual rollouts may be found in [eval_info.json](eval_info.json).
44
+
45
  <blank>|Ours
46
  -|-
47
+ Average max. overlap ratio for 500 episodes | 0.887
48
  Success rate for 500 episodes (%) | 66.0