The dataset used in this repo combines an official four_types dataset with depth information and additional self-generated datasets featuring variations in clothing colors and table colors.
The model is trained using ACT with the following training configuration:
batch_size: 32 | steps: 60000 | save_freq: 1000 | log_freq: 500
Training is conducted on an NVIDIA GeForce RTX 5090 GPU.
More info: [GitHub Link] (https://github.com/NYTCEE/r15-Lehome-Challenge)
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support