This model was trained to predict instructions from outputs as described in the "Self-Alignment with Instruction Backtranslation" paper.
The model was trained on the OpenAssistant-Guanaco dataset.
-
