hwanheelee1993 / UMIC

An unreferenced image captioning metric (ACL-21)
MIT License
29 stars 17 forks source link

Missing and mismatching files #1

Closed blmoistawinde closed 3 years ago

blmoistawinde commented 3 years ago

Hi Authors,

I greatly appreciate your work, establishing a new SOTA for referenced caption evaluation metric, and your contributions to open source. However, I noticed some issues in the current release.

It seems that some important files/directories for training the model and computing the score is not here by now, such as data/ and model/ as is in UNITER.

I also noticed some mismatches between README and the actual files, like launch_activate.sh, compute_score.py in README, while launch_container.sh and compute_metric.py in the current repo.

Any plan to fix them? Thanks.

hwanheelee1993 commented 3 years ago

I'm really sorry for the late release. I was making another version without using the docker(will be more easy to use) as stated in the README, which is different from the developed version. Thank you for waiting and the interest on our work. I'll fix it until this weekend.

hwanheelee1993 commented 3 years ago

I'm sorry for the late release. I fixed the code and made the new version without docker. Also, I provided the pre processed image feature, text feature for the dataset we used in the paper, and the code for preprocessing new files.