LaGeM: A Large Geometry Model for 3D Representation Learning and Diffusion (ICLR 2025)

OpenReview

arxiv

Project page

Training

16 GPUs (4 GPUs with accum_iter 4)

torchrun --nproc_per_node=4 main_ae.py \
        --accum_iter=4 \
        --model AutoEncoder \
        --output output/ae \
        --num_workers 32 --point_cloud_size 8192 \
        --batch_size 16 --epoch 500 \
        --warmup_epochs 1 --blr 5e-5 --clip_grad 1

Inference

python infer.py --pc_path your_point_cloud_saved_in_ply.ply

Pretrained Model

https://huggingface.co/Zbalpha/LaGeM

Bibtex

@inproceedings{
zhang2025lagem,
title={{LaGeM}: A Large Geometry Model for 3D Representation Learning and Diffusion},
author={Biao Zhang and Peter Wonka},
booktitle={The Thirteenth International Conference on Learning Representations},
year={2025},
url={https://openreview.net/forum?id=72OSO38a2z}
}