https://github.com/identxxy/tinn

Taichi implementation of tiny-cuda-nn
https://github.com/identxxy/tinn

Last synced: about 2 months ago
JSON representation

Taichi implementation of tiny-cuda-nn

• Host: GitHub
• URL: https://github.com/identxxy/tinn
• Owner: identxxy
• Created: 2022-11-28T13:19:20.000Z (about 2 years ago)
• Default Branch: master
• Last Pushed: 2022-12-11T23:19:38.000Z (about 2 years ago)
• Last Synced: 2024-07-08T11:33:00.219Z (6 months ago)
• Language: Python
• Size: 2.53 MB
• Stars: 6
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.MD

Awesome Lists containing this project

• awesome-taichi - tinn - A Taichi implementation of the tiny-cuda-nn. (Applications / **Machine Learning**)

README

        
# tinn

A library to create a large number of micro-scale neural networks using [Taichi](https://docs.taichi-lang.org/) language.  

Framework is inspired by [tiny-cuda-nn](https://github.com/NVlabs/tiny-cuda-nn).

## Feature

The **parrallelization** is among **all the networks**, rather than the matrix computation inside one NN, between layers.  

All the matrix computation between nerual network layers are **unrolled** by Taichi at compile time, executing in all parrallel threads,

which means the maxtrix size cannot be too large.  

When the matrix size exceed 32, e.g. for a 6-input 6-output layer, which means 36 matrix elements (>32), Taichi will give out a compile warning.  

## Demo

Network config is located at `data/*.json`

### MLP learn an image

```

python mlp_learn_an_image.py

```

For simplication, settings are located at the head of `mlp_learn_an_image.py`.  

#### Control

|key/mouse| visualied field | description |

|-|-|-|

`` | - |train all NN | 

click LMB | - | train the NN at cursor postion| 

`t` | train_batch | random xy coord [0, 1)|

`i` | inference_batch | meshgrid xy coord [0, 1) |

`g` | train_target | sampled pixel from reference |

`r` | reference_img | ground truth GT|

`l` | inference_loss | inference loss to GT |

`p` | - | print profiler info |

||||