-
Book Overview & Buying
-
Table Of Contents
Supercharged Coding with GenAI
By :
If the LLM does not answer questions well, there are four general courses of action: improving prompt engineering, adding more data through the context window, providing examples (few-shot learning), fine-tuning, and retrieval-augmented generation (RAG). These will be discussed in this section.
The discipline of prompt engineering has recently emerged as a human interface to the LLM. Users or programs can send commands to, ask questions of, and apply research reasoning through the LLM. The discipline involves applying specific wording to maximize the chance of correct outputs without producing erroneous or biased outputs, including hallucinations. The process of writing prompts is often an iterative process where it may require multiple rounds to achieve the desired output. Most of the prompts discussed in this book focus on a single task with specific and clear wording, which is considered good practice. The five S’...