Files
vllm/examples/pooling/score/colbert_rerank_online.py
Ilya Boytsov 439afa4eea feat: Add ColBERT late interaction model support (#33686)
Signed-off-by: Ilya Boytsov <ilyaboytsov1805@gmail.com>
Signed-off-by: Ilya Boytsov <boytsovpanamera@mail.ru>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
Co-authored-by: wang.yuqi <yuqi.wang@daocloud.io>
2026-02-05 08:05:13 +08:00

58 lines
1.7 KiB
Python

# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
"""
Example of using ColBERT late interaction model for reranking.
ColBERT (Contextualized Late Interaction over BERT) uses per-token embeddings
and MaxSim scoring for document reranking, providing better accuracy than
single-vector models while being more efficient than cross-encoders.
Start the server with:
vllm serve answerdotai/answerai-colbert-small-v1
Then run this script:
python colbert_rerank_online.py
"""
import json
import requests
url = "http://127.0.0.1:8000/rerank"
headers = {"accept": "application/json", "Content-Type": "application/json"}
data = {
"model": "answerdotai/answerai-colbert-small-v1",
"query": "What is machine learning?",
"documents": [
"Machine learning is a subset of artificial intelligence.",
"Python is a programming language.",
"Deep learning uses neural networks for complex tasks.",
"The weather today is sunny.",
],
}
def main():
response = requests.post(url, headers=headers, json=data)
if response.status_code == 200:
print("ColBERT Rerank Request successful!")
result = response.json()
print(json.dumps(result, indent=2))
# Show ranked results
print("\nRanked documents (most relevant first):")
for item in result["results"]:
doc_idx = item["index"]
score = item["relevance_score"]
print(f" Score {score:.4f}: {data['documents'][doc_idx]}")
else:
print(f"Request failed with status code: {response.status_code}")
print(response.text)
if __name__ == "__main__":
main()