langtech-bsc / argilla-eval

0 stars 0 forks source link

A/B test argilla tempalte

Clone this repository and cd.

git clone git@github.com:langtech-bsc/argilla-eval.git

INSTALLATION

Cd into the repository and download the docker-compose.yaml

wget -O docker-compose.yaml https://raw.githubusercontent.com/argilla-io/argilla/main/examples/deployments/docker/docker-compose.yaml

Deploy the Argilla server locally.

docker compose up -d

Create a venv and install requirements

python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

.ENV FILE

Create an .env file with the following variables.

Argilla server requires:

API_URL=http://localhost:6900
API_KEY=argilla.apikey

Depending on the ptoject change them:

RG_WORKSPACE=eval_workspace
NUMBER_USERS=3
DATASET_NAME=prompts-eval-dataset
DATASET_PATH=dataset.json

USAGE

Once Argilla server is deployed we need to interact with it via sdk.

Run workspace.py to create a workspace on the argilla server.

python workspace.py

Run users.py to handle users on workspaces.

python users.py

Generated dynamically from NUMBER_USERS. Users and passwords are sotred at root level asusers.csv

Run datasets.py to create and manipulate datasets.

python datasets.py

Datasets are loaded from /datasets/your_dataset.json

Run records to query and extract records.

python records.py

Rercods are stored at root level as records.json