Small LLMs vs. Fine-Tuned Bert for Classification: 32 Experiments (alex-jacobs.com)