Upload ACT model trained for 500k steps
Browse files- README.md +72 -0
- config.json +78 -0
- model.safetensors +3 -0
- policy_postprocessor.json +32 -0
- policy_postprocessor_step_0_unnormalizer_processor.safetensors +3 -0
- policy_preprocessor.json +72 -0
- policy_preprocessor_step_3_normalizer_processor.safetensors +3 -0
- train_config.json +201 -0
README.md
ADDED
|
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
tags:
|
| 4 |
+
- robotics
|
| 5 |
+
- act
|
| 6 |
+
- lerobot
|
| 7 |
+
- manipulation
|
| 8 |
+
- imitation-learning
|
| 9 |
+
datasets:
|
| 10 |
+
- gpudad/so101_pick_cube_chunked
|
| 11 |
+
library_name: lerobot
|
| 12 |
+
pipeline_tag: robotics
|
| 13 |
+
---
|
| 14 |
+
|
| 15 |
+
# ACT Model for SO-101 Pick Cube Task
|
| 16 |
+
|
| 17 |
+
This is an Action Chunking Transformer (ACT) model trained on the SO-101 robot arm for a cube picking task.
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
|
| 21 |
+
## Model Details
|
| 22 |
+
|
| 23 |
+
- **Architecture**: ACT (Action Chunking Transformer)
|
| 24 |
+
- **Vision Backbone**: ResNet18
|
| 25 |
+
- **Training Steps**: 500,000
|
| 26 |
+
- **Chunk Size**: 100
|
| 27 |
+
- **N Action Steps**: 1 (with temporal ensembling)
|
| 28 |
+
- **Temporal Ensemble Coefficient**: 0.01
|
| 29 |
+
- **KL Weight**: 10.0
|
| 30 |
+
- **Batch Size**: 16
|
| 31 |
+
- **Learning Rate**: 3e-5
|
| 32 |
+
|
| 33 |
+
## Training Dataset
|
| 34 |
+
|
| 35 |
+
Trained on [gpudad/so101_pick_cube_chunked](https://huggingface.co/datasets/gpudad/so101_pick_cube_chunked) - a chunked version of the SO-101 pick cube dataset with episode-level video files for efficient loading.
|
| 36 |
+
|
| 37 |
+
## Camera Views
|
| 38 |
+
|
| 39 |
+
The model uses 3 camera inputs:
|
| 40 |
+
- **Front camera** - Main observation view
|
| 41 |
+
- **Overhead camera** - Top-down perspective
|
| 42 |
+
- **Wrist camera** - End-effector mounted camera
|
| 43 |
+
|
| 44 |
+
## Training Command
|
| 45 |
+
|
| 46 |
+
```bash
|
| 47 |
+
python -m roboport.train act \
|
| 48 |
+
/root/datasets/so101_pick_cube_chunked \
|
| 49 |
+
-o /root/outputs/act_pick_cube_chunked \
|
| 50 |
+
--steps 500000 \
|
| 51 |
+
--chunk-size 100 \
|
| 52 |
+
--n-action-steps 1 \
|
| 53 |
+
--temporal-ensemble 0.01 \
|
| 54 |
+
--kl-weight 10.0 \
|
| 55 |
+
--batch-size 16 \
|
| 56 |
+
--lr 3e-5 \
|
| 57 |
+
--vision-backbone resnet18 \
|
| 58 |
+
--save-freq 50000 \
|
| 59 |
+
--gpu 0
|
| 60 |
+
```
|
| 61 |
+
|
| 62 |
+
## Usage
|
| 63 |
+
|
| 64 |
+
```python
|
| 65 |
+
from lerobot.common.policies.act.modeling_act import ACTPolicy
|
| 66 |
+
|
| 67 |
+
policy = ACTPolicy.from_pretrained("gpudad/act_so101_pick_cube")
|
| 68 |
+
```
|
| 69 |
+
|
| 70 |
+
## Framework
|
| 71 |
+
|
| 72 |
+
Trained using [roboport](https://github.com/DreamwareInc/roboport) with LeRobot backend.
|
config.json
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"type": "act",
|
| 3 |
+
"n_obs_steps": 1,
|
| 4 |
+
"input_features": {
|
| 5 |
+
"observation.state": {
|
| 6 |
+
"type": "STATE",
|
| 7 |
+
"shape": [
|
| 8 |
+
6
|
| 9 |
+
]
|
| 10 |
+
},
|
| 11 |
+
"observation.images.front": {
|
| 12 |
+
"type": "VISUAL",
|
| 13 |
+
"shape": [
|
| 14 |
+
3,
|
| 15 |
+
512,
|
| 16 |
+
512
|
| 17 |
+
]
|
| 18 |
+
},
|
| 19 |
+
"observation.images.wrist": {
|
| 20 |
+
"type": "VISUAL",
|
| 21 |
+
"shape": [
|
| 22 |
+
3,
|
| 23 |
+
512,
|
| 24 |
+
512
|
| 25 |
+
]
|
| 26 |
+
},
|
| 27 |
+
"observation.images.overhead": {
|
| 28 |
+
"type": "VISUAL",
|
| 29 |
+
"shape": [
|
| 30 |
+
3,
|
| 31 |
+
512,
|
| 32 |
+
512
|
| 33 |
+
]
|
| 34 |
+
}
|
| 35 |
+
},
|
| 36 |
+
"output_features": {
|
| 37 |
+
"action": {
|
| 38 |
+
"type": "ACTION",
|
| 39 |
+
"shape": [
|
| 40 |
+
6
|
| 41 |
+
]
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"device": "cuda",
|
| 45 |
+
"use_amp": true,
|
| 46 |
+
"push_to_hub": false,
|
| 47 |
+
"repo_id": null,
|
| 48 |
+
"private": null,
|
| 49 |
+
"tags": null,
|
| 50 |
+
"license": null,
|
| 51 |
+
"pretrained_path": null,
|
| 52 |
+
"chunk_size": 100,
|
| 53 |
+
"n_action_steps": 1,
|
| 54 |
+
"normalization_mapping": {
|
| 55 |
+
"VISUAL": "MEAN_STD",
|
| 56 |
+
"STATE": "MEAN_STD",
|
| 57 |
+
"ACTION": "MEAN_STD"
|
| 58 |
+
},
|
| 59 |
+
"vision_backbone": "resnet18",
|
| 60 |
+
"pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
|
| 61 |
+
"replace_final_stride_with_dilation": false,
|
| 62 |
+
"pre_norm": false,
|
| 63 |
+
"dim_model": 512,
|
| 64 |
+
"n_heads": 8,
|
| 65 |
+
"dim_feedforward": 3200,
|
| 66 |
+
"feedforward_activation": "relu",
|
| 67 |
+
"n_encoder_layers": 4,
|
| 68 |
+
"n_decoder_layers": 1,
|
| 69 |
+
"use_vae": true,
|
| 70 |
+
"latent_dim": 32,
|
| 71 |
+
"n_vae_encoder_layers": 4,
|
| 72 |
+
"temporal_ensemble_coeff": 0.01,
|
| 73 |
+
"dropout": 0.1,
|
| 74 |
+
"kl_weight": 10.0,
|
| 75 |
+
"optimizer_lr": 3e-05,
|
| 76 |
+
"optimizer_weight_decay": 0.0001,
|
| 77 |
+
"optimizer_lr_backbone": 1e-05
|
| 78 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:796d5dc247247aa01429b458a8beeabc4041db27588862ccc9b2a8a62670b0e3
|
| 3 |
+
size 206699736
|
policy_postprocessor.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"name": "policy_postprocessor",
|
| 3 |
+
"steps": [
|
| 4 |
+
{
|
| 5 |
+
"registry_name": "unnormalizer_processor",
|
| 6 |
+
"config": {
|
| 7 |
+
"eps": 1e-08,
|
| 8 |
+
"features": {
|
| 9 |
+
"action": {
|
| 10 |
+
"type": "ACTION",
|
| 11 |
+
"shape": [
|
| 12 |
+
6
|
| 13 |
+
]
|
| 14 |
+
}
|
| 15 |
+
},
|
| 16 |
+
"norm_map": {
|
| 17 |
+
"VISUAL": "MEAN_STD",
|
| 18 |
+
"STATE": "MEAN_STD",
|
| 19 |
+
"ACTION": "MEAN_STD"
|
| 20 |
+
}
|
| 21 |
+
},
|
| 22 |
+
"state_file": "policy_postprocessor_step_0_unnormalizer_processor.safetensors"
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"registry_name": "device_processor",
|
| 26 |
+
"config": {
|
| 27 |
+
"device": "cpu",
|
| 28 |
+
"float_dtype": null
|
| 29 |
+
}
|
| 30 |
+
}
|
| 31 |
+
]
|
| 32 |
+
}
|
policy_postprocessor_step_0_unnormalizer_processor.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a88726ee21decc4b8cd4f63b284a02c4f04437f77c9782708fc602ca60627166
|
| 3 |
+
size 2396
|
policy_preprocessor.json
ADDED
|
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"name": "policy_preprocessor",
|
| 3 |
+
"steps": [
|
| 4 |
+
{
|
| 5 |
+
"registry_name": "rename_observations_processor",
|
| 6 |
+
"config": {
|
| 7 |
+
"rename_map": {}
|
| 8 |
+
}
|
| 9 |
+
},
|
| 10 |
+
{
|
| 11 |
+
"registry_name": "to_batch_processor",
|
| 12 |
+
"config": {}
|
| 13 |
+
},
|
| 14 |
+
{
|
| 15 |
+
"registry_name": "device_processor",
|
| 16 |
+
"config": {
|
| 17 |
+
"device": "cuda",
|
| 18 |
+
"float_dtype": null
|
| 19 |
+
}
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"registry_name": "normalizer_processor",
|
| 23 |
+
"config": {
|
| 24 |
+
"eps": 1e-08,
|
| 25 |
+
"features": {
|
| 26 |
+
"observation.state": {
|
| 27 |
+
"type": "STATE",
|
| 28 |
+
"shape": [
|
| 29 |
+
6
|
| 30 |
+
]
|
| 31 |
+
},
|
| 32 |
+
"observation.images.front": {
|
| 33 |
+
"type": "VISUAL",
|
| 34 |
+
"shape": [
|
| 35 |
+
3,
|
| 36 |
+
512,
|
| 37 |
+
512
|
| 38 |
+
]
|
| 39 |
+
},
|
| 40 |
+
"observation.images.wrist": {
|
| 41 |
+
"type": "VISUAL",
|
| 42 |
+
"shape": [
|
| 43 |
+
3,
|
| 44 |
+
512,
|
| 45 |
+
512
|
| 46 |
+
]
|
| 47 |
+
},
|
| 48 |
+
"observation.images.overhead": {
|
| 49 |
+
"type": "VISUAL",
|
| 50 |
+
"shape": [
|
| 51 |
+
3,
|
| 52 |
+
512,
|
| 53 |
+
512
|
| 54 |
+
]
|
| 55 |
+
},
|
| 56 |
+
"action": {
|
| 57 |
+
"type": "ACTION",
|
| 58 |
+
"shape": [
|
| 59 |
+
6
|
| 60 |
+
]
|
| 61 |
+
}
|
| 62 |
+
},
|
| 63 |
+
"norm_map": {
|
| 64 |
+
"VISUAL": "MEAN_STD",
|
| 65 |
+
"STATE": "MEAN_STD",
|
| 66 |
+
"ACTION": "MEAN_STD"
|
| 67 |
+
}
|
| 68 |
+
},
|
| 69 |
+
"state_file": "policy_preprocessor_step_3_normalizer_processor.safetensors"
|
| 70 |
+
}
|
| 71 |
+
]
|
| 72 |
+
}
|
policy_preprocessor_step_3_normalizer_processor.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a88726ee21decc4b8cd4f63b284a02c4f04437f77c9782708fc602ca60627166
|
| 3 |
+
size 2396
|
train_config.json
ADDED
|
@@ -0,0 +1,201 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"dataset": {
|
| 3 |
+
"repo_id": "local/so101_pick_cube_chunked",
|
| 4 |
+
"root": "/root/datasets",
|
| 5 |
+
"episodes": null,
|
| 6 |
+
"image_transforms": {
|
| 7 |
+
"enable": false,
|
| 8 |
+
"max_num_transforms": 3,
|
| 9 |
+
"random_order": false,
|
| 10 |
+
"tfs": {
|
| 11 |
+
"brightness": {
|
| 12 |
+
"weight": 1.0,
|
| 13 |
+
"type": "ColorJitter",
|
| 14 |
+
"kwargs": {
|
| 15 |
+
"brightness": [
|
| 16 |
+
0.8,
|
| 17 |
+
1.2
|
| 18 |
+
]
|
| 19 |
+
}
|
| 20 |
+
},
|
| 21 |
+
"contrast": {
|
| 22 |
+
"weight": 1.0,
|
| 23 |
+
"type": "ColorJitter",
|
| 24 |
+
"kwargs": {
|
| 25 |
+
"contrast": [
|
| 26 |
+
0.8,
|
| 27 |
+
1.2
|
| 28 |
+
]
|
| 29 |
+
}
|
| 30 |
+
},
|
| 31 |
+
"saturation": {
|
| 32 |
+
"weight": 1.0,
|
| 33 |
+
"type": "ColorJitter",
|
| 34 |
+
"kwargs": {
|
| 35 |
+
"saturation": [
|
| 36 |
+
0.5,
|
| 37 |
+
1.5
|
| 38 |
+
]
|
| 39 |
+
}
|
| 40 |
+
},
|
| 41 |
+
"hue": {
|
| 42 |
+
"weight": 1.0,
|
| 43 |
+
"type": "ColorJitter",
|
| 44 |
+
"kwargs": {
|
| 45 |
+
"hue": [
|
| 46 |
+
-0.05,
|
| 47 |
+
0.05
|
| 48 |
+
]
|
| 49 |
+
}
|
| 50 |
+
},
|
| 51 |
+
"sharpness": {
|
| 52 |
+
"weight": 1.0,
|
| 53 |
+
"type": "SharpnessJitter",
|
| 54 |
+
"kwargs": {
|
| 55 |
+
"sharpness": [
|
| 56 |
+
0.5,
|
| 57 |
+
1.5
|
| 58 |
+
]
|
| 59 |
+
}
|
| 60 |
+
},
|
| 61 |
+
"affine": {
|
| 62 |
+
"weight": 1.0,
|
| 63 |
+
"type": "RandomAffine",
|
| 64 |
+
"kwargs": {
|
| 65 |
+
"degrees": [
|
| 66 |
+
-5.0,
|
| 67 |
+
5.0
|
| 68 |
+
],
|
| 69 |
+
"translate": [
|
| 70 |
+
0.05,
|
| 71 |
+
0.05
|
| 72 |
+
]
|
| 73 |
+
}
|
| 74 |
+
}
|
| 75 |
+
}
|
| 76 |
+
},
|
| 77 |
+
"revision": null,
|
| 78 |
+
"use_imagenet_stats": true,
|
| 79 |
+
"video_backend": "pyav",
|
| 80 |
+
"streaming": false
|
| 81 |
+
},
|
| 82 |
+
"env": null,
|
| 83 |
+
"policy": {
|
| 84 |
+
"type": "act",
|
| 85 |
+
"n_obs_steps": 1,
|
| 86 |
+
"input_features": {
|
| 87 |
+
"observation.state": {
|
| 88 |
+
"type": "STATE",
|
| 89 |
+
"shape": [
|
| 90 |
+
6
|
| 91 |
+
]
|
| 92 |
+
},
|
| 93 |
+
"observation.images.front": {
|
| 94 |
+
"type": "VISUAL",
|
| 95 |
+
"shape": [
|
| 96 |
+
3,
|
| 97 |
+
512,
|
| 98 |
+
512
|
| 99 |
+
]
|
| 100 |
+
},
|
| 101 |
+
"observation.images.wrist": {
|
| 102 |
+
"type": "VISUAL",
|
| 103 |
+
"shape": [
|
| 104 |
+
3,
|
| 105 |
+
512,
|
| 106 |
+
512
|
| 107 |
+
]
|
| 108 |
+
},
|
| 109 |
+
"observation.images.overhead": {
|
| 110 |
+
"type": "VISUAL",
|
| 111 |
+
"shape": [
|
| 112 |
+
3,
|
| 113 |
+
512,
|
| 114 |
+
512
|
| 115 |
+
]
|
| 116 |
+
}
|
| 117 |
+
},
|
| 118 |
+
"output_features": {
|
| 119 |
+
"action": {
|
| 120 |
+
"type": "ACTION",
|
| 121 |
+
"shape": [
|
| 122 |
+
6
|
| 123 |
+
]
|
| 124 |
+
}
|
| 125 |
+
},
|
| 126 |
+
"device": "cuda",
|
| 127 |
+
"use_amp": true,
|
| 128 |
+
"push_to_hub": false,
|
| 129 |
+
"repo_id": null,
|
| 130 |
+
"private": null,
|
| 131 |
+
"tags": null,
|
| 132 |
+
"license": null,
|
| 133 |
+
"pretrained_path": null,
|
| 134 |
+
"chunk_size": 100,
|
| 135 |
+
"n_action_steps": 1,
|
| 136 |
+
"normalization_mapping": {
|
| 137 |
+
"VISUAL": "MEAN_STD",
|
| 138 |
+
"STATE": "MEAN_STD",
|
| 139 |
+
"ACTION": "MEAN_STD"
|
| 140 |
+
},
|
| 141 |
+
"vision_backbone": "resnet18",
|
| 142 |
+
"pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
|
| 143 |
+
"replace_final_stride_with_dilation": false,
|
| 144 |
+
"pre_norm": false,
|
| 145 |
+
"dim_model": 512,
|
| 146 |
+
"n_heads": 8,
|
| 147 |
+
"dim_feedforward": 3200,
|
| 148 |
+
"feedforward_activation": "relu",
|
| 149 |
+
"n_encoder_layers": 4,
|
| 150 |
+
"n_decoder_layers": 1,
|
| 151 |
+
"use_vae": true,
|
| 152 |
+
"latent_dim": 32,
|
| 153 |
+
"n_vae_encoder_layers": 4,
|
| 154 |
+
"temporal_ensemble_coeff": 0.01,
|
| 155 |
+
"dropout": 0.1,
|
| 156 |
+
"kl_weight": 10.0,
|
| 157 |
+
"optimizer_lr": 3e-05,
|
| 158 |
+
"optimizer_weight_decay": 0.0001,
|
| 159 |
+
"optimizer_lr_backbone": 1e-05
|
| 160 |
+
},
|
| 161 |
+
"output_dir": "/root/outputs/act_pick_cube_chunked",
|
| 162 |
+
"job_name": "act",
|
| 163 |
+
"resume": false,
|
| 164 |
+
"seed": 1000,
|
| 165 |
+
"num_workers": 8,
|
| 166 |
+
"batch_size": 16,
|
| 167 |
+
"steps": 500000,
|
| 168 |
+
"eval_freq": 20000,
|
| 169 |
+
"log_freq": 200,
|
| 170 |
+
"save_checkpoint": true,
|
| 171 |
+
"save_freq": 50000,
|
| 172 |
+
"use_policy_training_preset": true,
|
| 173 |
+
"optimizer": {
|
| 174 |
+
"type": "adamw",
|
| 175 |
+
"lr": 3e-05,
|
| 176 |
+
"weight_decay": 0.0001,
|
| 177 |
+
"grad_clip_norm": 10.0,
|
| 178 |
+
"betas": [
|
| 179 |
+
0.9,
|
| 180 |
+
0.999
|
| 181 |
+
],
|
| 182 |
+
"eps": 1e-08
|
| 183 |
+
},
|
| 184 |
+
"scheduler": null,
|
| 185 |
+
"eval": {
|
| 186 |
+
"n_episodes": 50,
|
| 187 |
+
"batch_size": 50,
|
| 188 |
+
"use_async_envs": false
|
| 189 |
+
},
|
| 190 |
+
"wandb": {
|
| 191 |
+
"enable": false,
|
| 192 |
+
"disable_artifact": false,
|
| 193 |
+
"project": "roboport",
|
| 194 |
+
"entity": null,
|
| 195 |
+
"notes": null,
|
| 196 |
+
"run_id": null,
|
| 197 |
+
"mode": null
|
| 198 |
+
},
|
| 199 |
+
"checkpoint_path": null,
|
| 200 |
+
"rename_map": {}
|
| 201 |
+
}
|