We released our initial code for MCL, which requires explicit data augmentations. Getting Started Dependencies: python==3.6.13 pytorch==1.6.0. sentence-transformers==2.0.0. transformers==4.8.1. tensorboardX==2.4.1 pandas==1.1.5 sklearn==0.24.1 numpy==1.19.5

Step-1. download the original datastes from https://github.com/rashadulrakib/short-text-clustering-enhancement

step-2. then obtain the augmented data

step-3 run the code via the following:

python3 main.py
--resdir $path-to-store-your-results
--use_pretrain SBERT
--bert distilbert
--datapath $path-to-your-data
--dataname searchsnippets_trans_subst_20
--num_classes 8
--text text
--label label
--objective MCL
--augtype explicit
--temperature 0.5
--topk 500
--lr 1e-05
--lr_scale 100
--max_length 32
--batch_size 500
--max_iter 3000
--print_freq 100
--gpuid 0 &