Argument parsing with structured class.
Project description
ReRanker
ReRanker for hybrid retrieval.
For end-to-end text query service, check qtext.
Ranking
rank(query: str, docs: list[str])- cross-encoder model
- cohere model
- diversity (Max Marginal Relevance)
rank(docs: list[Record])- time decay with expressions
- title n-gram with bm25
- content n-gram with bm25
- document boost with expressions
- title embedding with content embedding
- title keywords with content keywords
- combination of the above features
Services
You can start all the service with the following command:
docker compose -f serving/compose.yaml up -d
Installation
pip install reranker
Usage
If you need the cross-encoder model service, you can start the server with the following command:
docker compose -f serving/compose.yaml up -d encoder
If you need the semantic highlight service, you can start the server with the following command:
docker compose -f serving/compose.yaml up -d colbert
from datetime import datetime, timedelta
from reranker import ReRanker, CrossEncoderClient, TimeDecayRanker
reranker = ReRanker(
steps=[
CrossEncoderClient(
model_name="cross-encoder/ms-marco-MiniLM-L-6-v2",
addr="http://localhost:8000",
),
TimeDecayRanker(decay_rate=0.5),
]
)
reranker.rank(
query="What is the capital of France?",
docs=[
"Paris is the capital of France.",
"The Eiffel Tower is in Paris.",
"The Louvre is in Paris.",
],
)
reranker.rank(
query=Record(
text="What is the capital of France?",
timestamp=datetime.now(),
),
docs=[
Record(
text="Paris is the capital of France.",
timestamp=datetime.now() - timedelta(days=1),
),
Record(
text="The Eiffel Tower is in Paris.",
timestamp=datetime.now() - timedelta(days=2),
),
Record(
text="The Louvre is in Paris.",
timestamp=datetime.now() - timedelta(days=3),
),
]
)
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file reranker-0.2.3.tar.gz.
File metadata
- Download URL: reranker-0.2.3.tar.gz
- Upload date:
- Size: 15.7 kB
- Tags: Source
- Uploaded using Trusted Publishing? Yes
- Uploaded via: twine/5.0.0 CPython/3.12.2
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
5e1ea0932d336b3181a6ecb4f511c1ed4bd5c907396b348fee7f28e0fee84da9
|
|
| MD5 |
3aea16e40dd18983ca929a99fc0bfc09
|
|
| BLAKE2b-256 |
21f4f58cda799752c0cecdcc6918d14117e13f2270c83351dd123a5bd13f1049
|
File details
Details for the file reranker-0.2.3-py3-none-any.whl.
File metadata
- Download URL: reranker-0.2.3-py3-none-any.whl
- Upload date:
- Size: 9.8 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? Yes
- Uploaded via: twine/5.0.0 CPython/3.12.2
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
86165605ef6a52b8de0f68dec80f3ca379b017f278215ad3aa1e5a9ac022f8e3
|
|
| MD5 |
38278e49a9989e580134d182b2cb3007
|
|
| BLAKE2b-256 |
88cd8291fe526e2776ab4dda2477d6d100ee3a02edcdb3f9b6f0d8a0419036c4
|