LlamaIndex Postprocessor Integration: TEI Rerank

Re-Rankers hosted on Text Embedding Inference Serve by Huggingface.

Install TEI Rerank package with: pip install llama-index-postprocessor-tei-rerank

text-embeddings-inference v0.4.0 added support for CamemBERT, RoBERTa and XLM-RoBERTa Sequence Classification models. Please refer to their repo for any further clarrification : https://github.com/huggingface/text-embeddings-inference

Docker start-up for TEI:

shell

model=BAAI/bge-reranker-large
volume=$PWD/data # share a volume with the Docker container to avoid downloading weights every run

docker run --gpus all -p 8080:80 -v $volume:/data --pull always ghcr.io/huggingface/text-embeddings-inference:1.5 --model-id $model --auto-truncate

Post successful startup of the docker image, the re-ranker can be initialised as follows:

python

from llama_index.postprocessor.tei_rerank import TextEmbeddingInference as TEIR

query_bundle = QueryBundle(prompt)
retrieved_nodes = retriever.retrieve(query_bundle)

postprocessor = TEIR(
    "BAAI/bge-reranker-large", "http://0.0.0.0/8080"
)  # Name of the model used in the docker server and base url (ip:port)

reranked_nodes = postprocessor.postprocess_nodes(
    nodes=retrieved_nodes, query_bundle=query_bundle
)