SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2109.04448 · EMNLP · 2021

Vision-and-Language or Vision-for-Language? On Cross-Modal Influence in Multimodal Transformers

Emanuele Bugliarello, Desmond Elliott, Stella Frank

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 7 functions out of this paper's own repositories and ran 1 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
e-bug/cross-modal-ablation canonical 0 of 6
ivana-13/guided_masking pwc_unofficial 1 of 1
FunctionStatusWhere it lives
get_index_positions Ran ivana-13/guided_masking/SVO_probes_word_clouds.py
code served (permissive licence) · get_code("fef294a1e5404f0c")
get_sentence_data Not yet run e-bug/cross-modal-ablation/volta/volta/datasets/flickr30ke_ablation_dataset.py
code served (permissive licence) · get_code("e0ae1e24559ddd25")
iot Not yet run e-bug/cross-modal-ablation/volta/volta/datasets/flickr30ke_ablation_dataset.py
code served (permissive licence) · get_code("a7ee1c19d2e17999")
iou Not yet run e-bug/cross-modal-ablation/volta/volta/datasets/flickr30ke_ablation_dataset.py
code served (permissive licence) · get_code("a1bd52099cee7c87")
kl_1601 Not yet run e-bug/cross-modal-ablation/volta/volta/losses.py
code served (permissive licence) · get_code("a2faafba33358747")
mse_2048 Not yet run e-bug/cross-modal-ablation/volta/volta/losses.py
code served (permissive licence) · get_code("1adf127958517c64")
nce_2048 Not yet run e-bug/cross-modal-ablation/volta/volta/losses.py
code served (permissive licence) · get_code("60cab169db54e986")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Pretrained vision-and-language BERTs aim to learn representations that combine information from both modalities. We propose a diagnostic method based on cross-modal input ablation to assess the extent to which these models actually integrate cross-modal information. This method involves ablating inputs from one modality, either entirely or selectively based on cross-modal grounding alignments, and evaluating the model prediction performance on the other modality. Model performance is measured by modality-specific tasks that mirror the model pretraining objectives (e.g. masked language modelling for text). Models that have learned to construct cross-modal representations using both modalities are expected to perform worse when inputs are missing from a modality. We find that recently proposed models have much greater relative difficulty predicting text when visual information is ablated, compared to predicting visual object categories when text is ablated, indicating that these models are not symmetrically cross-modal.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2109.04448")
get_code_for_paper("2109.04448")
have("2109.04448")

Connect an agent — have() is free.