#18 #diy #185h #4k #youtubeshorts #gjwpets #history #japan #dance #dance19 #xuhuong #beauty #nature #music #quansattru

1122
2 followers
#18 #diy #185h #4k #youtubeshorts #gjwpets #history #japan #dance #dance19 #xuhuong #beauty #nature #music #quansattru
Before we get into the architecture and into setting everything up,
I want to show you the result, basically what we're going to build.
And what we're going to start with is, we are going to start with CVs.
And I created with ChatGPT some a bit funny CVs from people that you might already know,
especially if you are a science fiction fan.
And these I filled with data engineer positions, like chief data engineer and so on.
And these CVs, we're going to extract the data and then create the embeddings for it.
And we're going to store everything into Elasticsearch as individual documents.
And these documents are going to have the text.
And they're going to have a vector where we're then running semantic search again,
so that we can find the right documents for our questions.
And how we're doing this on the interface is, we have a streamlined interface
where we can ask questions here.
We can filter by person, we can have a debug mode, where once we get an answer for our question,
we can then open this up and see which documents or which chunks
the LM then used to actually create us our result, our answer.
We're going to build this with Docker, we're going to build this with Elasticsearch
and Olama and Mistral LLMs.
So, pretty straightforward.
I'm going to talk about the architecture more,
but I want to just give you the idea or the overview of what are we going to build.
And you see, a chat interface, just like chat GPT or any other chatbot.
But in this, we're going to get the information from our rack system from Elasticsearch.
We're going to feed it into the LLM and then get good answers,
polished answers from it.
