---
license: gpl-3.0
---

The mft_model_ita.pth file contains the trained weights of a bert-based model fine-tuned for classifying text according to both the <b>moral dyad</b> discussed, i.e., one of

* 0:care/harm
* 1:fairness/cheating
* 2:loyalty/betrayal
* 3:authority/subversion
* 4:purity/degradation
* 5:no moral

and the <b>focus concern</b>, i.e, one of
* 0:prescriptive
* 1:prohibitive
* 2:no focus

These weights include the parameters of the custom layers, including the weights and biases of the two classifiers attached to the pre-trained model.

The model was built as part of the European project [VALAWAI](www.valawai.eu).