marzvrover / QuickExplain

QuickExplain

Quick Explain is a simple tool to explain complex topics.

Requirements

You can find python requirements in requirements.txt.

On Linux glibc version 2.32 or later is required.

You'll need at least 16gb of RAM for the default model. More/faster CPUs shoule mean faster response times.

Run

GitHub Codespaces

The postCreateCommand runs the .devcontainer/scripts/postcreate.sh script which installs python dependencies and downloads the default model, so you only need to run the sever.

./quickexplain.py

Locally

Install dependencies: pip install -r requirements.txt

Download LLM model: llm -m <model_name> "respond with: success" with default model: llm -m wizardlm-13b-v1 "respond with: success"

Run server: ./quickexplain.py

pip install -r requirements.txt
llm -m wizardlm-13b-v1 "respond with: success"
./quickexplain.py

Environment variables

Variable	Description	Default
QUICKEXPLAIN_DEBUG	Enable debug mode	False
QUICKEXPLAIN_MODEL	LLM model name	wizardlm-13b-v1
QUICKEXPLAIN_TOKEN_LIMIT	Maximum number of tokens to ingestible by model	2048
QUICKEXPLAIN_HOST	Host to listen on	localhost
QUICKEXPLAIN_PORT	Port to listen on	8080