The author also criticizes the quality and functionality of many frameworks, papers, and APIs that claim state-of-the-art performance, describing them as barebones and difficult to use. Despite being an AI enthusiast, the author expresses concern about the current state of the field, likening it to crypto fraud targeted at businesses and investors.
Key takeaways:
- The author expresses concern over the lack of validation for many models and finetunes claiming to be state-of-the-art on HuggingFace.
- There is skepticism about the legality and ethics of the private training datasets used for many of these models, as well as the potential for contamination with test data.
- Many frameworks, papers, and APIs claiming top performance seem to be lacking in substance and functionality upon closer inspection.
- The author, despite being an AI enthusiast, is getting a sense of potential fraud from the current state of affairs, likening it to the crypto wave targeting businesses and investors.