A simple weaviate demo used for thesis deduplication
Weaviate documentation can be found here.
Compose files included in this project contain different presets for weaviate install
- docker-compose.yml - default preset, big 768d english language model with no NER component.
- docker-compose-en.yml - small 384d english language model with no NER component.
- docker-compose-multilang.yml - small 384d mutilanguage language model with no NER component.
- docker-compose-big-ner.yml - big 768d english language model with NER component.
- docker-compose-small-ner.yml - small 384d english language model with NER component.
Install requirements:
pip install -r requirements.txt
Run docker compose of your choice
cd compose
docker-compose [-f filename] up -d
Prepare your dataset.json file.
Run the indexer
python3 index.py
Start the aplication
streamlit run streamlit_client.py