This repository is our submission to the ICLR Reproducibility Challenge and course project of Machine Learning(CS273A) at University of California, Irvine.

For more detail about our methodology and result, please refer to our report. As for source code of the original work, please refer to https://github.com/nyu-mll/GLUE-baselines.

The project depends on the GloVE and glue_data. Please download the GloVE data externally https://nlp.stanford.edu/projects/glove/ and also execute download_glue_data.py first.

Simply execute the shell script file locate in src/run_stuff.sh. If you have GPU, please execute src/gpu_run_stuff.sh

If necessary, please modify the relative data path