**GenTUS** is a user simulator for task-oriented dialogues, which consists of an encoder-decoder structure and is able to optimise both the user policy and natural language generation jointly. GenTUS generates both semantic actions and natural language utterances, preserving interpretability and enhancing language variation. In addition, by representing the inputs and outputs as word sequences and by using a large pre-trained language model we can achieve generalisability in feature representation.
**GenTUS** is a user simulator for task-oriented dialogues, which consists of an encoder-decoder structure and is able to optimise both the user policy and natural language generation jointly. GenTUS generates both semantic actions and natural language utterances, preserving interpretability and enhancing language variation. In addition, by representing the inputs and outputs as word sequences and by using a large pre-trained language model we can achieve generalisability in feature representation.
## Updates
* Use our latest version of GenTUS on ConvLab3
We recommend you to try GenTUS on ConvLab3! [GenTUS on ConvLab3](https://github.com/ConvLab/ConvLab-3/tree/master/convlab/policy/genTUS)!
The ConvLab3 repo is still under maintain so you can have the latest version.
In addition, we support unified-format, which means you can also try it various datasets, e.g. SGD and Taskmaster!
