---
library_name: lerobot
pipeline_tag: robotics
tags:
  - lerobot
  - robotics
  - pi05
datasets:
  - DreamMachines/arena2036_insertion_2op_224x224
base_model: lerobot/pi05_base
license: apache-2.0
---

# arena2036_insertion/pi05_2op

**π₀.₅ (Pi05)** — Vision-Language-Action model with open-world generalization, from Physical Intelligence. See the [π₀.₅ blog post](https://www.physicalintelligence.company/blog/pi05).

Trained with [LeDream](https://github.com/Dream-Machines-Robotics/dreammachines) (a fork of [LeRobot](https://github.com/huggingface/lerobot)).

## Training

- **Base model:** [`lerobot/pi05_base`](https://huggingface.co/lerobot/pi05_base)
- **Dataset:** [`DreamMachines/arena2036_insertion_2op_224x224`](https://huggingface.co/datasets/DreamMachines/arena2036_insertion_2op_224x224)
- **Step (this revision):** `30000`
- **Batch size:** `32`
- **Total steps target:** `30000`
- **Optimizer:** `adamw` (lr `0.0001`)

## Available revisions

Every training checkpoint is uploaded as its own git revision; the latest is also on `main`. Load a specific step with:

```python
from lerobot.policies.factory import make_policy
from lerobot.configs.policies import PreTrainedConfig

cfg = PreTrainedConfig.from_pretrained("DreamMachines/arena2036_insertion_pi05_2op", revision="v002000")
policy = make_policy(cfg)
```

Available: `v005000`, `v010000`, `v015000`, `v020000`, `v025000`, `v030000`

## Quick eval on a real DK1

```bash
make eval MODEL=DreamMachines/arena2036_insertion_pi05_2op
```

See [`docs/source/policy_pi05_README.md`](https://github.com/Dream-Machines-Robotics/dreammachines/blob/main/ledream/docs/source/policy_pi05_README.md) for π₀.₅-specific notes.