Question about train loader

[chongyangwang-song]

In StrongBaseLine， after generate pseudo labels， We need to update train loader According to the clustering results。 If only 4000 images were labeled， then the trainLoader is generated only for the 4000 images？ the remain images which were not labeled would not be used？ Thank you！