Hey, I want to create a custom LLM for resume parsing and grading and further integrate it with CrewAI. I researched about Sagemaker and Ollama but unsure about how and where to start. Custom-LLM is a requirement due to the safety and privacy concerns of enterprise data being used .
To create you own custom LLM it involves al lot of recourses and data, I think you have to use an already built LLM! There are a few free short courses that might help you with that in this site!
Also you can check the Generative AI with Large Language models in Coursera!
oh, well I am thinking about using GPT or Cohere and then maybe fine-tuning it or in a better case use a RAG pipeline for the LLM entirely? though I am still skeptical about data privacy due to GPT using the data to further train its models. Should i dive into federated learning for this?
Cannot help with federated learning, and cannot guarantee data privacy either with chatgpt! Unless it doesn’t transmit data over the web!
I will further look into isolated environments and RAG Pipelines for this! Also, yes GPT does that from my findings, will look for capable open-source models in HuggingFace for this! Thanks for helping out!