feat: make learning rate change when experiment resume

feat: make learning rate controllable by args

temp: fix resolution issue

temp: TTN resolution 256

chore: reduce learning rate

feat: change centorcrop same as original image size

feat: change TTN resolution from 256 to 512

fix: pretrain path does not editable

chore: increase logging interval

fix: remove redundant library

feat: add albumentizations library for transformation

chore: remove tqdm temporary

feat: change ttn transformation

feat: add test_root for test image log

docs: fix README