Brain-Score is a platform to evaluate computational models of vision and language on their match to behavioral and neural measurements in the domain of language processing. The intent of Brain-Score is to adopt many (ideally all) the experimental benchmarks in the field for the purpose of model testing, falsification, and comparison. To that end, Brain-Score operationalizes experimental data into quantitative benchmarks that any model candidate following the BrainModel interface can be scored on.
This Repo contains all the code needed to run and configure the website, including how to set up the site locally, the details of the website submission system, and both the front and back end infrastructure.
See the Developer Documentation for more details.
Brain-Score is made by and for the community. To contribute, please send in a pull request.