HUMAIN V1.0 is a Multi-Modal, Multi-Task chatbot project, empowered with 4 Generative AI models and was built on top of RAG-GPT and WebRAGQuery. Features:
Can act similar to ChatGPT
Has 3 RAG capabilities: RAG with processed docs, upload docs, and websites
Can generate images
Can summarize documents and websites
Connects a GPT model to the DuckDuckGo search engine (the model uses search engine functions automatically based on the user's query)
Can understand text, voice, and image.
Has built-in memory for the GPT models.
(The project is fully developed in Python)

🚀 GitHub Repository: github.com/Farzad-R/LLM-Zero-to-Hundred

00:00:00 Intro
00:01:19 Quick demo
00:06:36 Project requirements
00:06:36 Repository walk-through
00:12:12 Project schema
00:12:51 RAG-GPT and WebRAGQuery schema walk-through
00:16:30 Project structure
00:18:45 Process documents for RAG in advance
00:20:10 User interface design
00:23:45 RAG-GPT code explanation
00:33:00 Document summarization code explanation
00:36:39 RAG-GPT full demo
00:40:19 WebRAGQuery code explanation
00:58:40 Serving open-source Generative AI models (LLAVA, Diffusion, Whisper) code explanation
01:04:20 Multimodal full Demo
01:07:51 Final keynotes

Models:
GPT 3.5: OpenAI (Microsoft Azure)
text-embedding-ada-002: OpenAI (Microsoft Azure)
llava-hf/llava-v1.6-mistral-7b-hf: Higgingface
stabilityai/stable-diffusion-xl-base-1.0 : Higgingface
openai/whisper-base.en: Higgingface

HUMAIN has fully adapted RAG-GPT and WebRAGQuery projects. To watch those two projects please check the links below:
RAG-GPT:    • RAG-GPT: Chat with any documents and summa...  
WebRAGQuery:    • ChatGPT v2.0: Chat with Websites, Search t...  

The thumbnail's character is AI generated. Credit: Angelo Scarcella from Pixabay.

#openai #chatbot #multimodal #huggingface #diffusion #AI #generativeai #GPT #chatgpt #langchain #rag #LLM #largelanguagemodel #python #gradio