Skip to content

TrellixVulnTeam/simple_overfit_test_B1WY

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Overfitting Coordinate MLP to Colors

The objective of this repo is to simply overfit an MLP such that it takes coordinates as input and outputs colors. The data to overfit is provided in dataset/data. This data originally comes from a mesh with vertex colors.

Until now, I've been unsuccessful in achieving this objective.

File Structure

Folder Description
dataset/data Contains the mesh data the needs to be overfit
dataset/mesh_data.py Contains dataset class for training
model/net_texture.py MLP with positional encoding
trainer/train_overfit.py Training module
runs Experiment checkpoints and outputs

Requirements

Requirements are provided in requirements.txt file. To install them, run

pip install -r requirements.txt

Training

To train the MLP simply run

python trainer/train_overfit.py

About

Weirdly challenging Simple MLP overfitting problem

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%