Downloads · 30 days
0
roco_model_act_1
yjsm1203/roco_model_act_1
roco_model_act_1 is a robotics model from yjsm1203. Use it for the robotics task on the model card, and read the license before you ship it in a product. It is set up for act. The card lists the license as mit.
This model was trained using the ACT (Action Chunking with Transformers) policy for the Gearbox Assembly task.
Access
Public
Updated Dec 26, 2025
Repo size
5 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.ckpt5 GB · 100%
At a glance
- Task
- Robotics
- Library
- act
- License
- mit
- Access
- Public
- Created
- Dec 26, 2025
- Updated
- Dec 26, 2025
- SHA
- fb02c3ef
Model card
From the Hugging Face model README
ACT Model for Gearbox Assembly
This model was trained using the ACT (Action Chunking with Transformers) policy for the Gearbox Assembly task.
Model Configuration
The model was trained with the following hyperparameters:
- Task Name:
sim_gearbox_assembly_demos_filtered - Policy Class:
ACT - KL Weight: 10
- Chunk Size: 100
- Hidden Dim: 512
- Batch Size: 32
- Feedforward Dimension: 3200
- Number of Epochs: 10000
- Learning Rate: 1e-5
- Seed: 0
- Save Interval: 1000
- Early Stopping Patience: 3000
Training Command
nohup python3 /root/gearboxAssembly/source/Galaxea_Lab_External/Galaxea_Lab_External/VLA/ACT/act/imitate_episodes.py \
--task_name sim_gearbox_assembly_demos_filtered \
--ckpt_dir /root/gearboxAssembly/data/model \
--policy_class ACT --kl_weight 10 \
--chunk_size 100 \
--hidden_dim 512 \
--batch_size 32 \
--dim_feedforward 3200 \
--num_epochs 10000 \
--lr 1e-5 --seed 0 \
--save_interval 1000 \
--early_stopping_patience 3000 > training.log 2>&1 &
Checkpoints
The training process saved checkpoints every 1000 epochs.
Dataset
Trained on sim_gearbox_assembly_demos_filtered dataset path: /root/gearboxAssembly/data/model.
- Task
- Robotics
- Library
- act
- License
- mit
- Created
- Dec 26, 2025
- Updated
- Dec 26, 2025