chATLAS_Frontend 1.0.0

Code to run the frontend app

chATLAS

Welcome to the repository for the frontend of chATLAS: An AI assistant for the ATLAS collaboration

Main app (more stable): https://chatlas-flask-chatlas.app.cern.ch

Staging area (contains newer features, less stable): https://chatlas-staging-chatlas.app.cern.ch

If you want to install the app for development work, follow the instructions below:

Requirements

You will need to set the following environment variables for the app to run

export CHATLAS_OPENAI_KEY=... # Feel free to use a personal OpenAI API key here
export CHATLAS_DB_PASSWORD=...
export CHATLAS_EMBEDDING_MODEL_PATH=...
export CHATLAS_GROQ_KEY=...
export CHATLAS_GROQ_BASE_URL=http://cs-513-ml003:3000 # this will change if not on CERN network, see below

Reach out to the developers by email or on mattermost to get setup.

If running locally (not on lxplus), you'll need a local copy of the embedding model, see these instructions

Environment creation

Move to the directory containing this file and run:

uv sync

Running the app

0. Ensure you have the venv and necessary environment variables

export CHATLAS_OPENAI_KEY=...
export CHATLAS_DB_PASSWORD=...
export CHATLAS_EMBEDDING_MODEL_PATH=...
export CHATLAS_GROQ_KEY=...
export CHATLAS_PORT_FORWARDING=1

We recommend putting this into a .env file in the root directory of the project. Then you can load in the environment variables from the file with:

set -a
source .env
set +a

1. Setup port forwarding for the DBOD instances

note If you are connected to the CERN network, skip this step

# forward the all DBs
ssh -N \
  -L 6624:dbod-chatlas.cern.ch:6624 \
  -L 6606:dbod-chatlas-cds.cern.ch:6606 \
  -L 3000:cs-513-ml003:3000 \
  "$LXPLUS_USERNAME"@lxplus.cern.ch

2. Launch

uv run --env-file .env chATLAS_Frontend/launch.py --local-mode --db-host cern-prod --port-forwarding

3. Viewing the output

After launching, you should get some terminal output like " * Running on http://127.0.0.1:5000"

If running locally, just open this link

If on lxplus, you need to port-forward the address to your local machine. Note the lxplus node you are on and run:

ssh -L 5000:127.0.0.1:5000 <USERNAME>@lxplus<NUMBER>.cern.ch

Getting the Embedding Model

The embedding model is now stored on EOS. To run locally you need to download the embedding model:

scp <USERNAME>@lxplus.cern.ch:/eos/atlas/atlascerngroupdisk/phys-mlf/Chatlas/multi-qa-mpnet-base-dot-v1-ATLAS-TALK <LOCAL PATH TO STORE MODEL>

Then export environment variable:

export CHATLAS_EMBEDDING_MODEL_PATH = <LOCAL PATH TO STORE MODEL>

Updating the environment

Edit pyproject.toml and run uv sync to update the lockfile