Call for Papers

The BabyVLM Workshop invites original work at the underexplored intersection of multimodal machine learning, cognitive science, and developmental psychology. We welcome work and perspectives focused on questions where the interaction of language and other modalities is key for learning, or for evaluating the capabilities of sample-efficient language learners.

Submissions are invited under the following themes (which are not mutually exclusive):

  1. Developmentally plausible multimodal language models. Pretraining multimodal LMs using only as much data as a human has access to when first learning language, including the BabyVLM Challenge.

  2. Developmentally aligned evaluation. Evaluations and benchmarks inspired by human language and vision capabilities, measuring psychometric fit to human metrics and enabling a more fine-grained view of multimodal language learning during earlier stages of pretraining.

  3. Longitudinal egocentric learning. Studying and improving longitudinal, egocentric datasets to enable high-impact research.

Submission

To manage conflicts of interest, no organizer will be involved in assessing a submission from someone within the same organization, and no organizers nor any students with a conflict of interest with the organizers will submit papers to the workshop.

Important Dates

Code of Ethics and Conduct

All participants of the workshop (including authors and reviewers) are required to adhere to the NeurIPS Code of Ethics and NeurIPS Code of Conduct.