cross-encoder-RoBERTa-MarginMSE

This model is a cross-encoder based on FacebookAI/roberta-base. 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("xpmir/cross-encoder-RoBERTa-MarginMSE")
model = AutoModelForSequenceClassification.from_pretrained("xpmir/cross-encoder-RoBERTa-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	39.22	45.68
trec2019	93.90	70.72
trec2020	92.96	69.82
fever	81.73	81.45
arguana	23.79	34.91
climate_fever	34.38	25.58
dbpedia	77.42	46.76
fiqa	46.14	39.03
hotpotqa	90.21	74.70
nfcorpus	53.41	32.99
nq	55.03	59.99
quora	80.97	82.79
scidocs	28.34	15.85
scifact	67.37	69.75
touche	61.57	34.76
trec_covid	90.90	67.53
robust04	65.26	44.67
lotte_writing	69.96	60.77
lotte_recreation	62.96	57.49
lotte_science	48.33	40.27
lotte_technology	56.45	47.36
lotte_lifestyle	74.60	64.74
Mean In Domain	75.36	62.07
BEIR 13	60.87	51.24
LoTTE (OOD)	62.93	52.55