wirawan/eli5 repository overview

⁠The Open Domain Long Form Question and Answering System

This is an image for Open Domain Long Form Question and Answering System as explained at (https://yjernite.github.io/lfqa.html⁠). It is used currently as backend for the @UncoolAIBot⁠ chatbot on Telegram which is described at https://github.com/cahya-wirawan/AskMeAnything-Chatbot⁠.

⁠Usage

Create a directory /eli5/.cache before you run the docker image for the first time, It needs around 25GB disk space and run:

docker run --gpus all -p 8000:8000 -v /eli5/.cache:/root/.cache wirawan/eli5.

The first run will take more than 2 hours (depend on your network and server performance). All processed files are saved in local host file system /eli5/.cache, so we can use it for the following run without long pre-processing time.

⁠Server requirement

• A Windows/Linux server with docker, 100GB Disk space and 16GB RAM
• NVidia GPU (16GB RAM)