cross-encoder-bert-base-DistillRankNET

This model is a cross-encoder based on bert-base-uncased. It was trained on Ms-Marco using loss distillRankNET as part of a reproducibility paper for training cross encoders: "Reproducing and Comparing Distillation Techniques for Cross-Encoders", see the paper for more details.

Model Description
Usage
Evals

Model Description

This model is intended for re-ranking the top results returned by a retrieval system (like BM25, Bi-Encoders or SPLADE).

Training Data: MS MARCO Passage
Language: English
Loss distillRankNET

Training can be easily reproduced using the assiciated repository. The exact training configuration used for this model is also detailed in config.yaml.

Usage

Quick Start:

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
model = AutoModelForSequenceClassification.from_pretrained("xpmir/cross-encoder-bert-base-DistillRankNET")

features = tokenizer("What is experimaestro ?", "Experimaestro is a powerful framework for ML experiments management...", padding=True, truncation=True, return_tensors="pt")

model.eval()
with torch.no_grad():
    scores = model(**features).logits
    print(scores)

Evaluations

We provide evaluations of this cross-encoder re-ranking the top 1000 documents retrieved by naver/splade-v3-distilbert.

dataset	RR@10	nDCG@10
msmarco_dev	36.42	42.84
trec2019	95.74	74.15
trec2020	94.25	72.10
fever	81.04	80.99
arguana	22.80	34.31
climate_fever	29.17	21.50
dbpedia	76.58	45.80
fiqa	43.41	35.34
hotpotqa	89.45	72.86
nfcorpus	56.85	34.36
nq	52.57	57.27
quora	76.95	78.94
scidocs	28.31	15.65
scifact	67.81	70.21
touche	63.22	34.36
trec_covid	89.83	68.52
robust04	69.69	47.75
lotte_writing	64.88	55.85
lotte_recreation	58.11	52.84
lotte_science	43.32	36.06
lotte_technology	49.62	41.06
lotte_lifestyle	70.00	60.53
Mean In Domain	75.47	63.03
BEIR 13	59.85	50.01
LoTTE (OOD)	59.27	49.01