General Point Model with Autoencoding and Autoregressive

TL;DR

The paper introduces a versatile General Point Model (GPM) that combines autoencoding and autoregressive training for point cloud transformers, improving understanding and generation tasks with superior performance.

Contribution

It presents a novel unified framework integrating autoencoding and autoregressive tasks in point cloud transformers, enhancing performance and versatility.

Findings

GPM outperforms existing models like Point-BERT, MaskPoint, and PointMAE in understanding tasks.

GPM achieves competitive results in unconditional point cloud generation.

The model demonstrates potential for conditional generation by modifying input information.

Abstract

The pre-training architectures of large language models encompass various types, including autoencoding models, autoregressive models, and encoder-decoder models. We posit that any modality can potentially benefit from a large language model, as long as it undergoes vector quantization to become discrete tokens. Inspired by GLM, we propose a General Point Model (GPM) which seamlessly integrates autoencoding and autoregressive tasks in point cloud transformer. This model is versatile, allowing fine-tuning for downstream point cloud representation tasks, as well as unconditional and conditional generation tasks. GPM enhances masked prediction in autoencoding through various forms of mask padding tasks, leading to improved performance in point cloud understanding. Additionally, GPM demonstrates highly competitive results in unconditional point cloud generation tasks, even exhibiting the potential for conditional generation tasks by modifying the input's conditional information. Compared to models like Point-BERT, MaskPoint and PointMAE, our GPM achieves superior performance in point cloud understanding tasks. Furthermore, the integration of autoregressive and autoencoding within the same transformer underscores its versatility across different downstream tasks.