Closed andsteing closed 2 years ago
Code was tested with the following command:
big_vision.tools.eval_only --config big_vision/configs/proj/image_text/lit_coco.py:txt=bert_base,img=B/16_2,init=gs://vit_models/lit/LiT-B16B_2.npz
(to generate results added in configs/proj/image_text/README.md
in this PR)
Thanks! LGTM.
This is mirroring the changes in https://github.com/google-research/vision_transformer/pull/226