Our Objectives

With HuggingFace’s Transformers and Tokenizers libraries, many smaller businesses and individual researchers have become empowered to build their own Natural Language Processing (NLP) applications. It helps that (for the most part) the community itself around NLP research and application encourages an open-source exchange of ideas, tools, studies, even data. Many of the big players in AI Research and Development (such as Facebook, Microsoft, Google) have even shared their findings on the HuggingFace Hub, arXiv, among other platforms, which has enabled a lot of independent AI research to flourish.

At SolipsisAI, we aim to continue in this spirit of collaboration in the open exchange of ideas and data in our own projects. We will be posting our research notes, developer findings, and general development updates on this blog and prepare our internal projects for release as open-source libraries and applications. See our GitHub for our projects in progress.

The decentralization of NLP applications is another aspect we are focused on. These applications tend to be reliant on third-party resources and services that are often quite expensive to sustain. Furthermore, end users are forced to surrender their data to these providers. We are investigating ways to take these applications offline so that the data entered by an end user is theirs and theirs alone. This will be very challenging as most NLP apps can be resource-intensive, often requiring very powerful hardware (such as GPUs with massive VRAM capacity).

To that end, we hope our work in this space will help advance the field.