Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,48 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model:
|
| 4 |
+
- google/electra-base-discriminator
|
| 5 |
+
pipeline_tag: text-ranking
|
| 6 |
+
tags:
|
| 7 |
+
- cross-encoder
|
| 8 |
+
library_name: lightning-ir
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Set-Encoder: Permutation-Invariant Inter-Passage Attention for Listwise Passage Re-Ranking with Cross-Encoders
|
| 12 |
+
|
| 13 |
+
This model is presented in the paper [Set-Encoder: Permutation-Invariant Inter-Passage Attention for Listwise Passage Re-Ranking with Cross-Encoders](https://huggingface.co/papers/2404.06912). It's a cross-encoder architecture designed for efficient and permutation-invariant passage re-ranking.
|
| 14 |
+
|
| 15 |
+
Code: https://github.com/webis-de/set-encoder
|
| 16 |
+
|
| 17 |
+
We provide the following fine-tuned models for novelty-aware re-ranking.
|
| 18 |
+
|
| 19 |
+
To reproduce the results, run the following command using the [Lightning IR](https://github.com/webis-de/lightning-ir) library and the configuration files from the repository repository linked above:
|
| 20 |
+
|
| 21 |
+
```bash
|
| 22 |
+
lightning-ir re_rank --config ./configs/re-rank-novelty.yaml --model.model_name_or_path <MODEL_NAME>
|
| 23 |
+
```
|
| 24 |
+
|
| 25 |
+
(alpha nDCG@10, alpha=0.99 on TREC DL 19 and TREC DL 20)
|
| 26 |
+
|
| 27 |
+
| Model Name | TREC DL 19 (BM25) | TREC DL 20 (BM25) | TREC DL 19 (ColBERTv2) | TREC DL 20 (ColBERTv2) |
|
| 28 |
+
| ---------------------------------------------------------------------------------------- | ----------------- | ----------------- | ---------------------- | ---------------------- |
|
| 29 |
+
| [webis/set-encoder-novelty-base](https://huggingface.co/webis/set-encoder-novelty-base) | 0.805 | 0.721 | 0.821 | 0.803 |
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
## Citation
|
| 33 |
+
|
| 34 |
+
If you use this code or the models in your research, please cite our paper:
|
| 35 |
+
|
| 36 |
+
```bibtex
|
| 37 |
+
@InProceedings{schlatt:2025,
|
| 38 |
+
address = {Berlin Heidelberg New York},
|
| 39 |
+
author = {Ferdinand Schlatt and Maik Fr{\"o}be and Harrisen Scells and Shengyao Zhuang and Bevan Koopman and Guido Zuccon and Benno Stein and Martin Potthast and Matthias Hagen},
|
| 40 |
+
booktitle = {Advances in Information Retrieval. 47th European Conference on IR Research (ECIR 2025)},
|
| 41 |
+
doi = {10.1007/978-3-031-88711-6_1},
|
| 42 |
+
month = apr,
|
| 43 |
+
publisher = {Springer},
|
| 44 |
+
series = {Lecture Notes in Computer Science},
|
| 45 |
+
site = {Lucca, Italy},
|
| 46 |
+
title = {{Set-Encoder: Permutation-Invariant Inter-Passage Attention for Listwise Passage Re-Ranking with Cross-Encoders}},
|
| 47 |
+
year = 2025
|
| 48 |
+
}
|