Software developer and self-proclaimed spreadsheet aficionado Ishan Anand has accomplished the unthinkable by managing to jam the powerful GPT-2 language model into Microsoft Excel.
Anand’s astonishing achievement not only provides valuable insights into the inner workings of large language models (LLMs) and the underlying Transformer architecture but also serves as a testament to the boundless potential of human ingenuity.
“If you can understand a spreadsheet, then you can understand AI,” Anand confidently proclaims, inviting everyone to explore the depths of his 1.25GB spreadsheet marvel, which he has generously made available on GitHub.
While Anand’s spreadsheet implementation of GPT-2 may be a few steps behind the cutting-edge LLMs of 2024, it serves as a remarkable demonstration of the technology’s capabilities. By harnessing the power of smart next-token prediction, Anand’s creation can complete input sequences with uncanny accuracy, offering a tantalizing glimpse into the future of artificial intelligence.
Anand’s “low-code introduction” to LLMs is poised to be a game-changer, making this complex technology accessible to a wide range of professionals, from tech executives and marketers to policymakers, ethicists, and even developers and scientists new to the field of AI.
Also Read: OpenAI’s Murati Dodges Questions on Sora’s Training Data
