Views
No views yet
1 --dataset.repo_id=Tron-Ann/so101-transfer-cube \
2 --policy.type=pi0 \
3 --policy.pretrained_path=lerobot/pi0_base \
4 --policy.repo_id=Tron-Ann/pi0-so101-test \
5 --policy.dtype=bfloat16 \
6 --policy.gradient_checkpointing=true \
7 --output_dir=outputs/pi0_smoketest \
8 --job_name=pi0_smoketest \
9 --wandb.enable=false \
10 --steps=100 \
11 --batch_size=4 \
12 --policy.device=cuda```
13
14**π₀ (Pi0)**
15
16π₀ is a Vision-Language-Action model for general robot control, from Physical Intelligence. The LeRobot implementation is adapted from their open source OpenPI repository.
17
18**Model Overview**
19
20π₀ represents a breakthrough in robotics as the first general-purpose robot foundation model developed by Physical Intelligence. Unlike traditional robots that are narrow specialists programmed for repetitive motions, π₀ is designed to be a generalist policy that can understand visual inputs, interpret natural language instructions, and control a variety of different robots across diverse tasks.
21
22For more details, see the [Physical Intelligence π₀ blog post](https://www.physicalintelligence.company/blog/pi0).
23
24
25This policy has been trained and pushed to the Hub using [LeRobot](https://github.com/huggingface/lerobot).
26See the full documentation at [LeRobot Docs](https://huggingface.co/docs/lerobot/index).
27
28---
29
30## How to Get Started with the Model
31
32For a complete walkthrough, see the [training guide](https://huggingface.co/docs/lerobot/il_robots#train-a-policy).
33Below is the short version on how to train and run inference/eval:
34
35### Train from scratch
36
37```bash
38lerobot-train \
39 --dataset.repo_id=${HF_USER}/<dataset> \
40 --policy.type=act \
41 --output_dir=outputs/train/<desired_policy_repo_id> \
42 --job_name=lerobot_training \
43 --policy.device=cuda \
44 --policy.repo_id=${HF_USER}/<desired_policy_repo_id>
45 --wandb.enable=trueoutputs/train/<desired_policy_repo_id>/checkpoints/.1lerobot-record \
2 --robot.type=so100_follower \
3 --dataset.repo_id=<hf_user>/eval_<dataset> \
4 --policy.path=<hf_user>/<desired_policy_repo_id> \
5 --episodes=10--policy.path pointing to a local or hub checkpoint.