cross-encoder-RoBERTa-Hinge

This model is a cross-encoder based on FacebookAI/roberta-base. It was trained on Ms-Marco using loss hingeLoss 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 hingeLoss

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("FacebookAI/roberta-base")
model = AutoModelForSequenceClassification.from_pretrained("xpmir/cross-encoder-RoBERTa-Hinge")

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	37.36	43.79
trec2019	91.98	70.80
trec2020	92.96	69.29
fever	79.66	79.64
arguana	20.78	30.82
climate_fever	28.22	20.91
dbpedia	75.65	45.01
fiqa	46.61	38.75
hotpotqa	86.93	70.52
nfcorpus	51.61	30.94
nq	51.91	56.84
quora	75.27	77.80
scidocs	27.30	15.25
scifact	64.78	67.82
touche	60.31	32.87
trec_covid	91.07	70.13
robust04	66.00	44.23
lotte_writing	67.75	57.69
lotte_recreation	61.13	56.06
lotte_science	46.16	38.26
lotte_technology	53.46	44.13
lotte_lifestyle	72.64	63.23
Mean In Domain	74.10	61.29
BEIR 13	58.47	49.02
LoTTE (OOD)	61.19	50.60