Hello everyone! Welcome to all the new folks streaming into OpenAI and GPT-3 due to recent news! Many of you have questions and ideas about finetuning. I have been using finetuning since they released it, and done dozens of experiments, both with GPT-3 and other models. Let’s get right down to busi…

This will be so helpful for plenty of people here. Thanks Mate

Great content @daveshapautomator !

Hi. I’d love to see a tutorial demonstrating how to implement semantic search in an app! Thank you!

[image] daveshapautomator: Finetuning is for Structure, not Knowledge It’s worth mentioning that the confusion comes from the fact that GPT-3 has knowledge incorporated and the first thing one would think is that fine-tuning means to add more knowledge to it. It is possible to add knowledge …

This post saved a lot of days I would have spent trying to create a training data set with “prompts” of potential questions on some new knowledge and “completions” with answers to those questions. Now I’ll explore vector embeddings. Thank you.

[image] daveshapautomator: People that can’t get finetuning to work are often asking for orange juice from a cow. This one just summarize the whole dilemma about fine-tuning :slight_smile:

Semantic search is a useful tool, but this does not mean that there are no problems with semantic search, as with fine tuning. With a large number of documents, there is no guarantee that the tool will find the right document. It may happen that the information that the model needs for the answer is…

i’d wanna take advantage of the dialog management which chatgpt gives while semantic search does not. It saves a lot not to rebuild the context.

Ok, so for knowledge injection, embedding is the way, got it. But let’s take this use case as an example, code generation, embedded models cannot generate code for as long as i can remember, so i have to go through the models that can be fine tuned, then how do I effectively inject knowledge and co…

Finetuning for Domain Knowledge and Questions

Documentation

branchette February 13, 2023, 4:29pm 10

This article should go into the main openai documentation. I have been asking the question in a different thread: Fine-tune vs Embedding - #2 by ruby_coder

Topic		Replies	Views
How to fine tune a chatbot for Q&A API	11	8739	January 7, 2023
Fine-Tuning plus Embedding API	2	5045	May 3, 2023
What to do when fine-tuning is not working? API	20	8478	April 18, 2023
What does fine tuning actually do? (Fine tuning vs. Knowledge Retrieval) Documentation fine-tuning	7	7020	April 15, 2024
How to give domain knowledge to chatGPT API	4	7821	February 12, 2023

Finetuning for Domain Knowledge and Questions

Related topics