Thank you for sharing your excellent code!
I have some questions on the calculation of Purity metrics shown in Table 5 in the EMNLP'18 paper.
I could not find the part of purity in this code. When I try to implement by myself, I'm not sure about some details.
Could you please push the code of calculating purity?
Thank you!
Sorry for the delay; it should now be possible to calculate purity using the '-align' argument in chsmm.py, and the code should reproduce the purity numbers in the (updated) paper.
Thank you for sharing your excellent code! I have some questions on the calculation of Purity metrics shown in Table 5 in the EMNLP'18 paper. I could not find the part of purity in this code. When I try to implement by myself, I'm not sure about some details. Could you please push the code of calculating purity? Thank you!