Year: 2023
Type: conference-paper
Abstract: Evaluating grounded neural language model performance with respect to pragmatic qualities like the trade off between truthfulness, contrastivity and overinformativity of generated utterances remains a challenge in absence of data collected from humans.To enable such evaluation, we present a novel open source image-text dataset "Annotated 3D Shapes" (A3DS) comprising over nine million exhaustive natural language annotations and over 12 million variablegranularity captions for the 480,000 images provided by Burgess and Kim (2018).We showcase the evaluation of pragmatic abilities developed by a task-neutral image captioner fine-tuned in a multi-agent communication setting to produce contrastive captions.The evaluation is enabled by the dataset because the exhaustive annotations allow to quantify the presence of contrastive features in the model's generations.We show that the model develops human-like patterns (informativity, brevity, over-informativity for specific features (e.g., shape, color biases)).
Authors: Polina Tsvilodub, Michael Franke
Institution: University of Tübingen
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Funder: Deutsche Forschungsgemeinschaft
OpenAlex ID: W4385572250