Open-source LLMs: Uncensored & secure AI locally with RAG

ChatGPT is useful, but have you noticed that there are many censored topics, you are pushed in certain political directions, some harmless questions go unanswered, and our data might not be secure with OpenAI? This is where open-source LLMs like Llama3, Mistral, Grok, Falkon, Phi3, and Command R+ can help!

Are you ready to master the nuances of open-source LLMs and harness their full potential for various applications, from data analysis to creating chatbots and AI agents? Then this course is for you!

Introduction to Open-Source LLMs

This course provides a comprehensive introduction to the world of open-source LLMs. You'll learn about the differences between open-source and closed-source models and discover why open-source LLMs are an attractive alternative. Topics such as ChatGPT, Llama, and Mistral will be covered in detail. Additionally, you’ll learn about the available LLMs and how to choose the best models for your needs. The course places special emphasis on the disadvantages of closed-source LLMs and the pros and cons of open-source LLMs like Llama3 and Mistral.

Practical Application of Open-Source LLMs

The course guides you through the simplest way to run open-source LLMs locally and what you need for this setup. You will learn about the prerequisites, the installation of LM Studio, and alternative methods for operating LLMs. Furthermore, you will learn how to use open-source models in LM Studio, understand the difference between censored and uncensored LLMs, and explore various use cases. The course also covers finetuning an open-source model with Huggingface or Google Colab and using vision models for image recognition.

Prompt Engineering and Cloud Deployment

An important part of the course is prompt engineering for open-source LLMs. You will learn how to use HuggingChat as an interface, utilize system prompts in prompt engineering, and apply both basic and advanced prompt engineering techniques. The course also provides insights into creating your own assistants in HuggingChat and using open-source LLMs with fast LPU chips instead of GPUs.

Function Calling, RAG, and Vector Databases

Learn what function calling is in LLMs and how to implement vector databases, embedding models, and retrieval-augmented generation (RAG). The course shows you how to install Anything LLM, set up a local server, and create a RAG chatbot with Anything LLM and LM Studio. You will also learn to perform function calling with Llama 3 and Anything LLM, summarize data, store it, and visualize it with Python.

Optimization and AI Agents

For optimizing your RAG apps, you will receive tips on data preparation and efficient use of tools like LlamaIndex and LlamaParse. Additionally, you will be introduced to the world of AI agents. You will learn what AI agents are, what tools are available, and how to install and use Flowise locally with Node.js. The course also offers practical insights into creating an AI agent that generates Python code and documentation, as well as using function calling and internet access.

Additional Applications and Tips

Finally, the course introduces text-to-speech (TTS) with Google Colab and finetuning open-source LLMs with Google Colab. You will learn how to rent GPUs from providers like Runpod or Massed Compute if your local PC isn’t sufficient. Additionally, you will explore innovative tools like Microsoft Autogen and CrewAI and how to use LangChain for developing AI agents.

Harness the transformative power of open-source LLM technology to develop innovative solutions and expand your understanding of their diverse applications. Sign up today and start your journey to becoming an expert in the world of large language models!