Step 7: Training — ACT Command

Before running

• Environment: you need to first set up the environment and upload the dataset to the cloud GPU as described in Cloud GPU training environment setup. ACT comes with the basic LeRobot environment, so no extra installation is needed
• Dataset: --dataset.root=~/lerobot_my_dataset_shake_hands in the command points to the handshake dataset collected in step 6. If you are training your own task, replace it with your own dataset name
• Output directory: if --output_dir already exists, it will directly raise FileExistsError; use a new directory name, or add --resume=true to continue training
• You can check the curves on wandb at any time during training, see View real-time training curves on wandb

Reference documentation

https://github.com/huggingface/lerobot/blob/46e19ae579f80ce66211afafd1c3c649c569131f/docs/source/act.mdx

https://github.com/huggingface/lerobot/blob/main/src/lerobot/scripts/lerobot_train.py

https://github.com/huggingface/lerobot/blob/main/src/lerobot/configs/train.py

Why start with the ACT algorithm

ACT is the first model most recommended for training when playing with LeRobot. Its advantages are as follows:

• The model is very lightweight, with only eighty million learnable parameters

• Training converges very quickly, and inference is also very fast

• You can see results after training for one hour on a single GPU

• The ACT model itself is very small; the archive is about 200MB, which makes it very easy to store and transfer. The model archive produced by training is about 300MB (see the end of this page)

• Collecting about 30 episodes of data for the dataset is basically enough

• It can be deployed for inference on an Ubuntu host, a Mac computer, a Windows computer, or even a Raspberry Pi

• The inference results on a real robot are still quite good, which is enough for simple tasks like grabbing, shaking hands, and placing a pen

• The ACT algorithm is already included in the basic environment of the LeRobot library, so no other libraries need to be installed

Command line

lerobot-train \
  --dataset.repo_id=<username>/lerobot_my_dataset_shake_hands \
  --dataset.root=~/lerobot_my_dataset_shake_hands \
  --dataset.revision=v0.1.0 \
  --dataset.streaming=false \
  --policy.type=act \
  --output_dir=~/output_lerobot_train/shake/act/ \
  --job_name=shake_act_a \
  --policy.device=cuda \
  --wandb.enable=true \
  --wandb.project=Lerobot_my_Project \
  --policy.push_to_hub=false \
  --steps=20000 \
  --batch_size=8

Command line reference

The line-continuation character `` can only have one space before it and no space after it

Command line parameter	Description
--dataset.repo_id	The Repo_ID of the HuggingFace dataset, in the form 用户名/数据集名
--dataset.root	The local path of the dataset. When the dataset has already been downloaded locally, it should point to the actual directory
--dataset.revision	The dataset version, which was specified when uploading the dataset to HuggingFace
--dataset.streaming	Whether to read in a streaming manner. When the dataset is local, set this to false, as streaming reads are not needed
--policy.type	The algorithm to train, such as act, smolvla, diffusion, pi0, pi05, pi0_fast, wall_x
--output_dir	The directory where the training output is saved
--job_name	The name of this training job
--policy.device	The compute device
--wandb.enable	Enable wandb visualization
--wandb.project	The wandb project name
--policy.push_to_hub	Upload the trained model to the HuggingFace cloud
--steps	The number of training steps
--batch_size	The amount of data fed in per step; if VRAM is insufficient, it should be reduced

Training process

The model archive is about 300MB

Related Products

SO-ARM101 Developer Kit→