Artifacts from the paper "Semantic reconstruction of continuous
language from non-invasive brain recordings" [[paper](https://www.nature.com/articles/s41593-023-01304-9)] [[code](https://github.com/HuthLab/semantic-decoding)].

This repo is organized as follows:
1. `data_lm/`: lanugage model for various tasks
2. `models/`: encoder model + word rate model for various tasks
3. `data_test/`: test data with (word, fmri) pairs for various tasks
4. `data_train/`: train data with (word, fmri) pairs for various subjects + tasks