Update README.md
Browse files
README.md
CHANGED
|
@@ -14,8 +14,6 @@ See the [LeRobot library](https://github.com/huggingface/lerobot) (particularly
|
|
| 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
|
|
@@ -26,6 +24,8 @@ python lerobot/scripts/train.py \
|
|
| 26 |
device=cuda
|
| 27 |
```
|
| 28 |
|
|
|
|
|
|
|
| 29 |
## Model Size
|
| 30 |
|
| 31 |
<blank>|Number of Parameters
|
|
@@ -40,9 +40,11 @@ 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 |
-
|
| 44 |
|
| 45 |
<blank>|Ours
|
| 46 |
-|-
|
| 47 |
Average max. overlap ratio for 500 episodes | 0.887
|
| 48 |
Success rate for 500 episodes (%) | 66.0
|
|
|
|
|
|
|
|
|
| 14 |
|
| 15 |
## Training Details
|
| 16 |
|
|
|
|
|
|
|
| 17 |
The model was trained using this command:
|
| 18 |
|
| 19 |
```bash
|
|
|
|
| 24 |
device=cuda
|
| 25 |
```
|
| 26 |
|
| 27 |
+
This took about 7 hours to train on an Nvida A6000.
|
| 28 |
+
|
| 29 |
## Model Size
|
| 30 |
|
| 31 |
<blank>|Number of Parameters
|
|
|
|
| 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 |
+
Here are the metrics for 500 episodes worth of evaluation.
|
| 44 |
|
| 45 |
<blank>|Ours
|
| 46 |
-|-
|
| 47 |
Average max. overlap ratio for 500 episodes | 0.887
|
| 48 |
Success rate for 500 episodes (%) | 66.0
|
| 49 |
+
|
| 50 |
+
The results of each of the individual rollouts may be found in [eval_info.json](eval_info.json).
|