/hyperbolic_llm

Primary LanguagePythonApache License 2.0Apache-2.0

Code for Hyperbolic Pre-Trained Language Model (TASLP).

Pre-Training

The code is written based on Nvidia's Deep Learning Example. You can refer to the original repo to see the guidelines on data preparation. We provide the script for pre-training hyperbolic BERT in scripts/run_bert.sh

Fine-Tuning

The fine-tuning scripts are provided in scripts.

Pre-Trained Model

We provide the pre-trained hyperbolic BERT here, you can download and extract to results/