genius-chatbot 1.2.0

Use huggingface models to create an intelligent and scalable chatbot

Genius Chatbot

Version: 1.2.0

Chatbot that uses any desired hugging face model or allows for scalable intelligence based on hardware limitations

Usage:

Short Flag	Long Flag	Description
-h	--help	See Usage
-a	--assimilate	Assimilate knowledge from media provided in directory
	--batch-token	Number of tokens per batch
	--chromadb-directory	Number of chunks to use
	--chunks	Number of chunks to use
-e	--embeddings-model	Embeddings model to use
	--hide-source	Hide source of answer
-j	--json	Export to JSON
	--openai-token	OpenAI Token
	--openai-api	OpenAI API Url
	--pgvector-user	PGVector user
	--pgvector-password	PGVector password
	--pgvector-host	PGVector host
	--pgvector-port	PGVector port
	--pgvector-database	PGVector database
	--pgvector-driver	PGVector driver
-p	--prompt	Prompt for chatbot
	--mute-stream	Mute stream of generation
-m	--model	Copy GPT4All .bin file from the Model Explorer
	--max-token-limit	Maximum token to generate
	--model-directory	Directory to store models locally
	--model-engine	GPT4All LlamaCPP, or OpenAI

Example:

Your very first time running should assimilate up to 3 documents to establish the Chroma Database. This will unlock prompting

genius-chatbot --assimilate "/directory/of/documents"

genius-chatbot --prompt "What is the 10th digit of Pi?"

genius-chatbot --prompt "Chatbots are cool because they" \
    --model "wizardlm-13b-v1.1-superhot-8k.ggmlv3.q4_0.bin" \
    --model-engine "GPT4All" \
    --assimilate "/directory/of/documents" \
    --json

Install Instructions

Install Python Package

Windows Prerequisites:

Visual Studio Code 2022

winget install -e --id Kitware.CMake

Ubuntu Prerequisites:

apt install -y pandoc

python -m pip install genius-chatbot

Build Instructions

Build Python Package

sudo chmod +x ./*.py
sudo pip install .
python3 setup.py bdist_wheel --universal
# Test Pypi
twine upload --repository-url https://test.pypi.org/legacy/ dist/* --verbose -u "Username" -p "Password"
# Prod Pypi
twine upload dist/* --verbose -u "Username" -p "Password"