MY FINETUNED MODELS BEAT OPENAI'S GPT-4

Summary from the TLDR Web Dev newsletter

This author evaluated several fine-tuned LLMs on structured data extraction from press releases and found that they outperformed OpenAI's GPT models. The evaluation process was challenging due to the different locations and interfaces of the models. While the author was pleased with the results, they also acknowledge the tradeoffs of fine-tuning LLMs, such as the need for managing data and model versions.