Popular repositories Loading

1. number-token-loss number-token-loss Public

   PyPI package for Number Token Loss (ICML 2025)

   Python 24 1

2. multiscale-byte-lm multiscale-byte-lm Public

   A hierarchical LM that scales to training on context windows of +5M tokens

   Python 10

3. dataset-explorer dataset-explorer Public

   Interactive Dataset Explorer App

   Python 1

4. textcat textcat Public

   This repo contains the code for training language models based on SMILES strings to represent adsorption structures. This work employs the Open Catalyst OC20 dataset.

   Jupyter Notebook 1