GAUDI: A Neural Architect for Immersive 3D Scene Generation
Authors: Miguel Angel Bautista, Pengsheng Guo, Samira Abnar, Walter Talbott, Alexander Toshev, Zhuoyuan Chen, Laurent Dinh, Shuangfei Zhai, Hanlin Goh, Daniel Ulbricht, Afshin Dehghan, Joshua Susskind
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section we show the applicability of GAUDI to multiple problems. First, we evaluate reconstruction quality and performance of the reconstruction stage. Then, we evaluate the performance of our model in generative tasks including unconditional and conditional inference, in which radiance fields are generated from conditioning variables corresponding to images or text prompts. Full experimental settings and details can be found in the appendix B. |
| Researcher Affiliation | Industry | Miguel Angel Bautista Pengsheng Guo Samira Abnar Walter Talbott Alexander Toshev Zhuoyuan Chen Laurent Dinh Shuangfei Zhai Hanlin Goh Daniel Ulbricht Afshin Dehghan Josh Susskind Apple https://github.com/apple/ml-gaudi Corresponding email: mbautistamartin@apple.com |
| Pseudocode | No | The paper describes the model architecture and processes but does not include any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | Yes | Apple https://github.com/apple/ml-gaudi |
| Open Datasets | Yes | We report results on 4 datasets: Vizdoom [23], Replica [64], VLN-CE [26] and ARKit Scenes [1], which vary in number of scenes and complexity (see Fig. 3 and Tab. 1). |
| Dataset Splits | No | The paper states that 'Full experimental settings and details can be found in the appendix B' and mentions 'sample 10 random images per trajectory to compute the reconstruction metrics', but it does not explicitly define training, validation, or test data splits with percentages or sample counts in the provided main text. |
| Hardware Specification | No | The paper states that 'These details will be clarified in B' regarding compute and resource types, but no specific hardware models (e.g., GPU/CPU types, memory) are mentioned in the provided text. |
| Software Dependencies | No | The paper does not explicitly list software dependencies with specific version numbers (e.g., Python 3.8, PyTorch 1.9) in the provided text. |
| Experiment Setup | Yes | For all our experiments we set the dimension of zscene and zpose to 2048 and β = 0.1 unless otherwise stated. Full experimental settings and details can be found in the appendix B. |