README.md

---
title: 📝GPT🔊4OSB🔬🧠
emoji: 📝🔊🔬🧠
colorFrom: red
colorTo: green
sdk: streamlit
sdk_version: 1.46.1
app_file: app.py
pinned: false
license: mit
---
# 7/9 - evaluate new GPT models

This experimental multi agent mixture of expert system uses a variety of techniques and models to create different combinatorial AI solutions.

Models Used:

1. Mistral-7B-Instruct
2. Llama2-7B
3. Mixtral-8x7B-Instruct
4. Google Gemma-7B
5. OpenAI Whisper Small En
6. OpenAI GPT-4o, Whisper-1
7. ArXiV Embeddings

The techniques below which are not ML models but AI include:
1. Speech Synthesis using browser technology
2. Memory for semantic facts, and episodic emotional and event time series memories
3. Web integration using the q= standard for search linking allowing comparison of tech giant AI implementations:
4. Bing then Bing copilot with click 2
5. Google which does an AI search now
6. Twitter, the new home for technology discoveries, AI Output and Grok
7. Wikipedia for fact checking
8. YouTube
9. File and metadata integration combining text, audio, image, and video

This app also merges common theories in cognitive AI, AI with python libraries (e.g. NLTK, SKLearn).

The intent is to demonstrate SOTA AI/ML and combinations of Function-Input-Output for interoperability and knowledge management.

This space also serves as an experimental test bed for new technologies mixing it in with old for comparison and integration.


--Aaron