Cool man thanks. So when you change the embedding can you still process it with ChatGPT? I mean, is it only used for internal use and storage? Or can we reuse it somehow with a new prompt?
This is for RAG systems! ChatGPT doesn’t really use this but other systems using GPT as APIs do. You use these embeddings to compare them and look for content in, for example, a vector database. Then once you find the information you want, you use the associated text to feed to a LLM like GPT-4! So yes in short it’s for storage and retrieval only, but kind of used in prompt in a way since it is associated with the text that was embedded too! :)
Nice work. Would love to see more such shorts.
Working on that! 😎
Cool man thanks. So when you change the embedding can you still process it with ChatGPT? I mean, is it only used for internal use and storage? Or can we reuse it somehow with a new prompt?
This is for RAG systems! ChatGPT doesn’t really use this but other systems using GPT as APIs do. You use these embeddings to compare them and look for content in, for example, a vector database. Then once you find the information you want, you use the associated text to feed to a LLM like GPT-4! So yes in short it’s for storage and retrieval only, but kind of used in prompt in a way since it is associated with the text that was embedded too! :)