ExtraSmall
Artificial Intelligence
Text Classification Challenge with Extra-Small Datasets: Positive-Tuning Versus ChatGPT The dataset Regular fine-tuning with RoBERTa Few-shot with ChatGPT Positive-tuning a GPT-3 model Conclusion Sources
LLMs excel on extra-small datasets, but classical approaches shine as datasets growOnce more, performance was heavily influenced by the prompt and the samples provided. The model also generated several categories outside the goal list,...
Recent posts
