Possible World Research is dedicated to researching and developing tiny AI models in the area of Natural Language Processing. Why tiny? Because we believe that tomorrow's technology should not pollute our beautiful planet. Also because small models can be used by everybody, irrespective of their financial resources. All our code is tested on consumer hardware: typically, home laptops and single board computers. No GPUs involved.

Developing tiny models means creating more intelligent algorithms. For any given task, a model that achieves high performance with fewer resources is arguably more intelligent. We believe that simulating human faculties in machines requires a good understanding of different scientific fields, including linguistics, psychology, and of course computer science. We welcome contributions that demonstrate some curiosity about the human mind.