Skip to content
No description, website, or topics provided.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.


Code style: black

This repository contains code for evaluating compositional generalization in image captioning models. It additionally contains code for two state-of-the-art image captioning models and a joint model for caption generation and image-sentence ranking, called BUTR. All models are implemented in PyTorch.

It accompanies the following CoNLL 2019 paper:

Compositional Generalization in Image Captioning
Mitja Nikolaus, Mostafa Abdou, Matthew Lamm, Rahul Aralikatte and Desmond Elliott


Show, Attend and Tell

An implementation of the Show, Attend and Tell model (adapted from a PyTorch Tutorial to Image Captioning).

Bottom-Up and Top-Down Attention

An implementation of the decoder of the Bottom-Up and Top-Down Attention model. Encoded features from the bottom-up attention model for the COCO dataset can be found on the project's GitHub page.

Bottom-Up and Top-Down attention with Re-ranking (BUTR)

An implementation of the joint model for caption generation and image-sentence ranking based on the Bottom-Up and Top-Down Attention and VSE++ models.

Compositional Generalization Evaluation

A model has to be trained and evaluated using the four different dataset splits. Afterwards, the resulting evaluation json files should be merged into a single json containing the results for all 24 held out concept pairs. The average recall@5 and some other statistics can be visualized using

You can’t perform that action at this time.