The Star Map for Navigating the GenAI Universe

GenAI is growing by double exponential. Navigating this space itself requires invention.

Selection or choice is one of the most valuable traits of Generative AI industry. However, the exponential growth of new models, capabilities, use cases, datasets, tools, and research makes this selection a super difficult task for any business. GenAI Techne Star Map is our endeavor to make this selection easily available, meaningfully searchable, and valuably relevant for your business needs.

Mapping the Universe of GenAI

As on Feb 2024, there are more than 500,000 models and 110,000 datasets on Hugging Face. These models and datasets are classified by more than 40 inference tasks like, Text-to-Image, Text Classification, and Image-to-Video. Each of these models has several attributes to determine if they are fit for a use case.

There are more than 2.4 million research paper submissions on arXiv. Computer Science is trending up growing from 4.5K submissions per month a year ago to 7.5K monthly.

A search for “LLM” on GitHub results in more than 35K repositories. Top 10 repos have more than 20K stars rating. Many of LLM or GenAI related open source projects have shown exponential growth.

Let us understand how expansive is this GenAI universe. We want to effectively evaluate one or many of 500,000 models using one or many of 100,000 datasets, across 40 odd inference tasks, which may be optimized by a selection of 35,000 tools… you get the idea. If we simply multiple the numbers in round brackets in the following partial star map, it results in 4.704e+26 or septillion scale permutations! To add to the complexity, the green highlighted parts of the star maps are growing every day!

Watch this space as we explore the GenAI Techne Start Map with a few examples, release tools to make this star map available for your business, and deliver insights as we process the GenAI universe.