Thanks to visit codestin.com
Credit goes to Github.com

Skip to content

giglr/nano-trm

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

67 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

An implementation of TRM

This is an implementation of the Tiny Recursive Model (TRM)

Reference code

Data

ARC-AGI: You need to download the data from the kaggle challenge page.

Other datasets are generated locally.

To generate a Sudoku dataset: ./bash/generate_sudoku_data.sh

Training

uv run python src/nn/train.py experiment=trm_sudoku4x4 (takes a few mins on an A10)

Evaluation

uv run python src/nn/evaluate.py +checkpoint=/tmp/ml-experiments/lunar-pine-174/checkpoints/last.ckpt +data_dir=./data/sudoku_4x4_small

Visualization

To visualize the results of a model, use notebooks/neural_viewer.ipynb

Comments / contributions

Follow me on X

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 74.1%
  • Jupyter Notebook 25.3%
  • Shell 0.6%