cross-encoder-bert-base-MarginMSE

This model is a cross-encoder based on bert-base-uncased. It was trained on Ms-Marco using loss marginMSE 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 marginMSE

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-MarginMSE")

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	40.10	46.59
trec2019	97.09	75.46
trec2020	95.52	74.02
fever	82.57	82.29
arguana	24.22	35.90
climate_fever	34.78	26.03
dbpedia	77.70	46.68
fiqa	45.25	37.40
hotpotqa	90.30	74.98
nfcorpus	57.07	34.34
nq	54.08	59.11
quora	77.81	80.10
scidocs	27.26	15.62
scifact	67.75	69.98
touche	59.28	34.82
trec_covid	92.13	66.43
robust04	71.18	48.23
lotte_writing	68.09	58.80
lotte_recreation	61.26	55.80
lotte_science	46.27	38.24
lotte_technology	53.52	44.86
lotte_lifestyle	71.50	62.34
Mean In Domain	77.57	65.36
BEIR 13	60.78	51.05
LoTTE (OOD)	61.97	51.38