This is an implementation of the paper: Deep Learning Meets Projective Clustering. (ICLR 2021)

Researchers:
    Alaa Malouf
    Harry Lang
    Daniela Rus
    Dan Feldman
    
Link to the paper: 
	https://openreview.net/pdf?id=EQfpYwF3-b

Main file: 
	src/run_exp.py 

Usage example:
python3  src/run_exp.py --model_name_or_path textattack/distilbert-base-uncased-RTE  --task_name RTE --do_eval --data_dir $PATHTOGLUEDATA/glue_data/RTE/ --max_seq_length 128 --per_device_train_batch_size 32 --learning_rate 2e-5  --no_cuda   --overwrite_output_dir --output_dir /tmp/rte_distil  --do_train --num_train_epochs 0


Comments:
	In our code we use the huggingface library to build our new architecture, where we modify some of its files to obtain a sparse layer after fine tuning.
	

