File size: 1,565 Bytes
1824b1c | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 | ---
library_name: openpi
tags:
- pi0.5
- pi05
- so101
- lerobot
- robotics
- lora
license: apache-2.0
---
# pi05-so101-lora-50demos
LoRA fine-tune of [pi05_base](https://huggingface.co/openpi/pi05_base) on
[jakegonz/pick-and-place-red-block-50demos](https://huggingface.co/datasets/jakegonz/pick-and-place-red-block-50demos).
| Step | Folder |
|------|--------|
| 5000 | `step_5000/` |
| 10000 | `step_10000/` |
| 15000 | `step_15000/` |
## Training config
- Base: `gs://openpi-assets/checkpoints/pi05_base/params`
- Train config: `pi05_so101_lora` (PaliGemma 2B LoRA + Gemma 300M LoRA action expert)
- Action horizon: 10 (≈0.33 s @ 30 Hz)
- Batch size: 32
- LR schedule: cosine, peak 5e-5, warmup 500 steps, decay over 10k
- Optimizer: AdamW with grad clip 1.0
## Inference inputs (per step)
```text
observation.state : float32 (6,) joint pos (5 arm + 1 gripper, 0=open/100=closed)
observation.images.camera1 : uint8 wrist camera → maps to `left_wrist_0_rgb`
observation.images.camera2 : uint8 overhead camera → maps to `base_0_rgb`
prompt : "Pick up the red block and place it"
```
## Loading
```python
from openpi.policies import policy_config
from openpi.training import config as _config
cfg = _config.get_config("pi05_so101_lora")
policy = policy_config.create_trained_policy(cfg, "step_15000")
```
Each step folder contains `params/`, `assets/`, and `_CHECKPOINT_METADATA` — the minimum
required by `create_trained_policy`. The `train_state/` optimizer state is not included.
|