In this session you will learn how to enhance ChatGPT with context specific knowledge. A set of custom documents will be used to extract relevant texts for the user query. These texts are provided as โpromptsโ to ChatGPT. Using this approach, ChatGPT can give more concise answers to questions in a specific domain without the need of expensive fine-tuning.
๐After the session, complete the following tutorial so you can earn points for the amazing Devtoberfest contest: https://developers.sap.com/tutorials/devtoberfest2023-week3-data-ai-qa-llms.html
Thank you Karim and @noravonthenen for the informative session on RAG and the insightful use of Langchain. I'm eagerly looking forward to accessing the Git repository and learning the steps to create a knowledge base and embedding. Your presentation was engaging and valuable!
the repo is available now:
https://github.com/karimmohraz/rag-movieplots
In order to improve the answers I am looking to use a model like mistral.
If you have a ChatGPT key you could also use that as "llm", which can be fed into langchatin.
Here is a post on how to deploy LLaMa2 on the SAP AI Core:
https://blogs.sap.com/2023/07/28/running-language-models-deploy-llama2-7b-on-ai-core/comment-page-1/...