the relationship between model extracting features and clustering result

[Di Qiu]

hi, here is a question about the mapping relation between model extracting features and clustering result, will the better model extracting features result in the better clustering fscore?

[Lei Yang]

Hi @qiudi0127 , thanks for the question. I think it is similar to #50.

Basically, it is very likely that better feature extractor would lead to better clustering results. However, as all clustering algorithms can benefit from stronger feature extractor, it is uncertain that the performance gap between our method and baseline methods will enlarge continuously. More experiments are required to show the change of performance gap.

[Di Qiu]

@yl-1993 hi, i met an issue that using the part1_test.bin and pretrained model can not get the fscore given by git.

[Lei Yang]

Hi @qiudi0127 , it seems that the k is set to 15 instead of 80 for testing?