Acquire and improve skills
to build and grow products

The fundamental difference between LLM applications and traditional machine learning is that in most cases, you do not tune the model’s parameters and hyperparameters. Instead, you tweak your prompt to fix errors and improve the model’s performance on your intended task. Without a systematic approach to analyzing errors and making corrections, you can get caught up in making random changes to your prompt without knowing how they affect the overall performance of your LLM application.