BabyLM Shared Task

evaluation-pipeline-2023. Evaluation pipeline for the BabyLM Challenge 2023.

evaluation-pipeline-2024. The evaluation pipeline for the 2024 BabyLM Challenge.

evaluation-pipeline-2025. Python

baseline-pretraining. Code for pre-training BabyLM baseline models.

babylm.github.io. HTML

babylm_data_preprocessing. Python

submissions2023. This repository distributes full model outputs for the BabyLM 2023 evaluation pipeline as well as scores per task and subtask for a subset of the submissions to the 2023 BabyLM Challenge. All outputs and scores are scraped from DynaBench. We additionally include links to published models.