How to integrtae Pandas AI with local LLM using Ollama? | Private & Free | Ollama | AI | llama3

hello and uh welcome one and all we finally have the pandas AI integration with local llms this is a really exciting news Panda's AI is a python library that Bridges the gap between Panda's data frames and large language models allowing us to interact with our data using natural language we can explore clean analyze our data using large language models with the recent updates and new open source model this is becoming a useful tool we'll use llama 3 a powerful open-source large language model from meta it is capable of Performing various tasks like summarization question answering and even code generation we will use AMA for the local large language model integration with Panda's AI I have covered AMA setup in this video here this is a prerequisite for today's session make sure you have Ama set up and you can pull down llama 3 with a simple command okay let's get back to this project we will create a virtual environment for this project so we'll issue the following command to create the environment once it is created we'll go ahead and activate it we'll start by installing the requirements the requirements are in the requirement text so we'll issue the following command pip install d r requirement. text this will install the required libraries so for this project we need pandas pandas AI Open PI Exel and we'll utilize chain lit later on to put this into a UI our required libraries are installed I'll go ahead and copy the data set it's a sales data and I'll copy it in the base directory this will be available in the GitHub repo we will carry out the coding in jupyter Notebook so I'll go ahead and create a new Jupiter notebook in this notebook we import the require libraries now we are going to go ahead and read our data into a data frame using pandas we are ready to integrate the large language model for this we'll use ama's API make sure AMA server is up and running we'll use theama 3 model for this we provide our data frame and the llm to the Panda's API with this our setup is complete and we are ready to analyze our data using Panda's AI let's ask ask you a basic question what is the total sales that is What's the total sales in this data set I'll go ahead and bring up the terminal where AMA is running we can see the realtime interaction with the llm I am going to open up another terminal so we can watch the utilization and performance of the GPU you enter the following command watch dasn 0.5 nvidia-smi this will monitor the gpus utilization let's go ahead and ask another question about our data set our Command is submitted by the llm and we see a spike in the GPU usage this is in real time as we see llm process the request via GPU this is another way of saying I've upgraded my GPU to get better performance from the llm anyways here's our answer let's try and get the Panda's API to produce a chart we submit our question the llm is processing our request the visualization takes a bit longer than the data request we can see the GPU is processing this request and voila the llm does its magic and Panda's API produces a bar chart so this is a bar chart of sales by product category this is simply amazing I am amazed at the speed this technology is not perfect there were few errors along the way but the model produced the results on the second or third try especially the visualization bit anyways this is the panda AI integration with local llm I hope you enjoyed the session like share and subscribe take care and I'll see you in the next video

Channel: BI Insights Inc

Views: 720

Keywords: pandasai, pandas, pandas ai, data science, machine learning, deep learning, artificial intelligence, data analysis, ai, generative ai, generative ai python, generative ai tutorial, llama 3, llama 380, llama 3 local, llama 3 8b, llama 3 vs gpt 4, llama 3 ai, llama 3 meta, llama 3 llm, ollama, ollama windows, ollama rag, ollama langchain, ollama docker, ollama api, ollama agents, llama ai, ollama basics, ollama beginner, ollama best models, data analysis using ai

Id: vCoUZBhYzHA

Length: 4min 14sec (254 seconds)

Published: Sun May 19 2024