Pinned Loading

1. sad-training sad-training Public

   The code and data used to train multimodal models with synthetic audio data, as described in Soubki and Murzaku et al. (2025), which was accepted to NAACL 2025 Findings.

   Python 2 1

2. cogstates/multimodal-belief cogstates/multimodal-belief Public

   The code and data used to train multimodal (and unimodal) belief prediction models in Murzaku, Soubki, and Rambow (2024), which was accepted to Interspeech 2024.

   Python

3. cogstates/common-tom cogstates/common-tom Public

   The code and data used to generate Common-ToM, a theory of mind (ToM) question answering corpus based on common ground.

   Python 1

4. cogstates/wikiface cogstates/wikiface Public

   The code and data used to train face act prediction models for Wikipedia talk pages in Soubki, Choi, and Rambow (2024), which was accepted to SIGDIAL 2024.

   Python