cross-encoder-DeBERTav3-BCE

This model is a cross-encoder based on microsoft/deberta-v3-base. It was trained on Ms-Marco using loss bce as part of a reproducibility paper for training cross encoders: "Reproducing and Comparing Distillation Techniques for Cross-Encoders", see the paper for more details.

Contents

• 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 bce

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("microsoft/deberta-v3-base")
model = AutoModelForSequenceClassification.from_pretrained("xpmir/cross-encoder-DeBERTav3-BCE")

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	36.37	42.91
trec2019	90.47	66.67
trec2020	90.19	64.49
fever	66.75	69.05
arguana	13.92	20.99
climate_fever	17.30	13.61
dbpedia	57.37	32.56
fiqa	43.10	36.33
hotpotqa	77.04	61.33
nfcorpus	38.25	21.67
nq	46.54	51.74
quora	47.44	51.12
scidocs	25.03	14.22
scifact	63.63	66.40
touche	56.30	30.33
trec_covid	89.38	72.78
robust04	55.79	35.44
lotte_writing	64.83	56.43
lotte_recreation	59.42	54.49
lotte_science	44.35	36.87
lotte_technology	51.93	44.39
lotte_lifestyle	73.10	64.26
Mean In Domain	72.34	58.02
BEIR 13	49.39	41.70
LoTTE (OOD)	58.24	48.65