Summary: Against LLM maximalism · Explosion

The new LLM support in spaCy now lets you plug in LLM-powered components for these prediction tasks, which is especially great for prototyping.

Large Language Models (LLMs) can be used for arbitrary prediction tasks, by constructing a prompt describing the task, giving the labels to predict, and optionally including a relatively small number of examples in the prompt.

Most practical tasks don’t require powerful reasoning abilities or extensive background world knowledge, which are the things that really set LLMs apart from smaller models.

Here’s how I think LLMs should be used in NLP projects today — an approach I would call LLM pragmatism.

It will run much more quickly and accurately than a chain of LLM calls, and you’ll know that no matter what text is passed in, your predictive components will always give you valid output.

Similar Articles

Against LLM maximalism · Explosion

LLMs are not a direct solution to most of the NLP use-cases companies have been working on. They are extremely useful, but if you want to deliver reliable software you can improve over time, you can’t just write a prompt and call it a day. Once you’re past prototyping and want to deliver the best system you can, supervised learning will often give you better efficiency, accuracy and reliability.

Read the complete article at: explosion.ai