***Demo.** All projects are required to have a demo. It’s open ended, but we provide some ideas on how to build demos in the [How to build a demo](#how-to-build-a-demo) section.
***Technical difficulty.** Difficulty has different aspects, such as working with complex architectures, obtaining better evaluation metrics than existing models, or implementing models for low-resource languages.
***Social impact.** The project is expected to have a positive social impact, e.g. by tackling under-explored area of practical interest for minorities or under-represented group (low-ressources languages, specific focus on bias, fairness or ethical issues in ML) or by tackling general societal challenges, e.g. health or climate related challenges.
***Innovativeness.** Projects that propose novel applications or bring new ideas will be rewarded more.
### Jury
*[Niki Parmar](https://research.google/people/NikiParmar/): Staff Research Scientist at Google.
