https://github.com/probcomp/bayes3d
https://github.com/probcomp/bayes3d
Last synced: 11 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/probcomp/bayes3d
- Owner: probcomp
- License: apache-2.0
- Created: 2022-10-06T17:53:12.000Z (over 3 years ago)
- Default Branch: main
- Last Pushed: 2025-05-05T20:28:53.000Z (11 months ago)
- Last Synced: 2025-05-08T22:56:53.273Z (11 months ago)
- Language: Jupyter Notebook
- Homepage: https://probcomp.github.io/bayes3d/
- Size: 571 MB
- Stars: 26
- Watchers: 6
- Forks: 6
- Open Issues: 9
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
![logo][logo]
[![License][license]][license-url]
![Stability: Experimental][experimental-badge]
- **[Documentation](https://probcomp.github.io/bayes3d/)**
- **[Installation](#installation-guide)**
- **[Get Assets](#get-assets)**
- **[Google Cloud Instance Setup](#gcp-setup)**
# Installation Guide
Set up a fresh Python environment:
```bash
conda create -n bayes3d python=3.9
conda activate bayes3d
```
Install compatible versions JAX and Torch:
```bash
pip install --upgrade torch==2.2.0 torchvision==0.17.0+cu118 --index-url https://download.pytorch.org/whl/cu118
pip install --upgrade jax[cuda11_local]==0.4.20 -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
```
Install Bayes3D:
```bash
pip install bayes3d
```
Download model and data assets:
```bash
wget -q -O - https://raw.githubusercontent.com/probcomp/bayes3d/main/download.sh | bash
```
## Test
Run `python demo.py` to test installation setup.
## Common issues
Error:
```
fatal error: EGL/egl.h: No such file or directory
#include
fatal error: GL/glu.h: No such file or directory
#include
```
Run:
```
sudo apt-get install mesa-common-dev libegl1-mesa-dev libglfw3-dev libgl1-mesa-dev libglu1-mesa-dev
```
Error:
```
[F glutil.cpp:338] eglInitialize() failed
Aborted (core dumped)
```
Reinstall NVIDIA drivers with `sudo apt-get install nvidia-driver-XXX`. Check version of driver using `nvidia-smi`.
Error:
```
ImportError: libcupti.so.11.7: cannot open shared object file: No such file or directory
```
Run:
```
pip install torch==2.0.0+cu118 torchvision==0.15.1+cu118 torchaudio==2.0.1 --index-url https://download.pytorch.org/whl/cu118
```
Error:
```
raise RuntimeError("Ninja is required to load C++ extensions")
```
Run:
```
sudo apt-get update
sudo apt-get install ninja-build
```
To check your CUDA version:
```
nvcc --version
```
# GCP Setup
- Start new VM instance (see
[link](https://cloud.google.com/compute/docs/instances/create-start-instance)).
Select GPU - NVIDIA V100 and Machine Type 8vCPU 4 Core 30GB.
-From the VM instances page, searched for public image `c2-deeplearning-pytorch-2-0-gpu-v20230925-debian-11-py310`. Increase storage to 1000GB.
- Note that public image names get updated frequently, so it is possible you may not find the one mentioned above. To find the latest public image, go to the [public list](https://cloud.google.com/compute/docs/images#console), and look for an image as close to the one above (Debian 11, CUDA 11.8, Python 3.10, Pytorch 2.0 etc.).
- SSH into instance and when prompted, install the NVIDIA drivers.
- Follow [installation guide](#installation-guide).
## License
Distributed under the [Apache 2.0](LICENSE) license. See [LICENSE](LICENSE).
[experimental-badge]: https://img.shields.io/badge/stability-experimental-orange.svg
[license-url]: LICENSE
[license]: https://img.shields.io/badge/License-Apache_2.0-brightgreen.svg
[logo]: https://github.com/probcomp/bayes3d/assets/66085644/bf4e3d42-2d70-40fa-b980-04bd4e18bf2b