Upload folder using huggingface_hub
Browse files- README.md +58 -0
- act.bin +3 -0
- act.xml +0 -0
- config.json +18 -0
- manifest.json +18 -0
README.md
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: physicalai-train
|
| 3 |
+
tags:
|
| 4 |
+
- vision-language-action
|
| 5 |
+
- robotics
|
| 6 |
+
- physicalai
|
| 7 |
+
license: apache-2.0
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Action Chunking Transformer
|
| 11 |
+
|
| 12 |
+
That's a basic model for solving simplest imitation learning tasks.
|
| 13 |
+
The original implementations can be found [here](https://github.com/tonyzhaozh/act/tree/main).
|
| 14 |
+
|
| 15 |
+
The model takes images from one or multiple cameras and robot state and produces
|
| 16 |
+
a chunk of actions, which robot can execute as a sequence of movements in real world.
|
| 17 |
+
|
| 18 |
+
The model weights are random and provided only for testing purposes.
|
| 19 |
+
|
| 20 |
+
# How to Use
|
| 21 |
+
|
| 22 |
+
## Installation
|
| 23 |
+
|
| 24 |
+
```bash
|
| 25 |
+
uv pip install physicalai-train
|
| 26 |
+
```
|
| 27 |
+
|
| 28 |
+
## Running Inference
|
| 29 |
+
|
| 30 |
+
The following API example showcases inference API for this model:
|
| 31 |
+
|
| 32 |
+
```python
|
| 33 |
+
import numpy as np
|
| 34 |
+
from physicalai.inference import InferenceModel
|
| 35 |
+
|
| 36 |
+
model = InferenceModel("act_ov_fp16", device="CPU")
|
| 37 |
+
|
| 38 |
+
# Build a dummy LIBERO-style observation.
|
| 39 |
+
# LIBERO provides two cameras (agentview + wrist) and an 8-dim robot state.
|
| 40 |
+
# Images use the LeRobot convention: float32 in [0, 1], shape (C, H, W).
|
| 41 |
+
observation = {
|
| 42 |
+
"images.image": np.random.rand(3, 256, 256).astype(np.float32),
|
| 43 |
+
"images.image2": np.random.rand(3, 256, 256).astype(np.float32),
|
| 44 |
+
"state": np.zeros(8, dtype=np.float32),
|
| 45 |
+
}
|
| 46 |
+
|
| 47 |
+
chunk = model.predict_action_chunk(observation)
|
| 48 |
+
```
|
| 49 |
+
|
| 50 |
+
Note, that the model should be downloaded and saved to `act_ov_fp16` folder prior to running this script.
|
| 51 |
+
|
| 52 |
+
## Legal information
|
| 53 |
+
|
| 54 |
+
The original model is distributed under [Apache 2.0](https://choosealicense.com/licenses/apache-2.0/) license.
|
| 55 |
+
|
| 56 |
+
## Disclaimer
|
| 57 |
+
|
| 58 |
+
Intel is committed to respecting human rights and avoiding causing or contributing to adverse impacts on human rights. See [Intel’s Global Human Rights Principles](https://www.intel.com/content/dam/www/central-libraries/us/en/documents/policy-human-rights.pdf). Intel’s products and software are intended only to be used in applications that do not cause or contribute to adverse impacts on human rights.
|
act.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7afc453ddc902249bf220fb90b031cc048cc2017b64df515124e12017fae625a
|
| 3 |
+
size 68412798
|
act.xml
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"policy": {
|
| 3 |
+
"name": "act",
|
| 4 |
+
"source": {
|
| 5 |
+
"class_path": "physicalai.policies.act.policy.ACT"
|
| 6 |
+
}
|
| 7 |
+
},
|
| 8 |
+
"model": {
|
| 9 |
+
"runner": {
|
| 10 |
+
"class_path": "physicalai.inference.runners.single_pass.SinglePass"
|
| 11 |
+
},
|
| 12 |
+
"artifacts": {
|
| 13 |
+
"openvino": "act.xml"
|
| 14 |
+
}
|
| 15 |
+
},
|
| 16 |
+
"format": "policy_package",
|
| 17 |
+
"version": "1.0"
|
| 18 |
+
}
|
manifest.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"policy": {
|
| 3 |
+
"name": "act",
|
| 4 |
+
"source": {
|
| 5 |
+
"class_path": "physicalai.policies.act.policy.ACT"
|
| 6 |
+
}
|
| 7 |
+
},
|
| 8 |
+
"model": {
|
| 9 |
+
"runner": {
|
| 10 |
+
"class_path": "physicalai.inference.runners.single_pass.SinglePass"
|
| 11 |
+
},
|
| 12 |
+
"artifacts": {
|
| 13 |
+
"openvino": "act.xml"
|
| 14 |
+
}
|
| 15 |
+
},
|
| 16 |
+
"format": "policy_package",
|
| 17 |
+
"version": "1.0"
|
| 18 |
+
}
|