cross-encoder-DeBERTav3-infoNCE

This model is a cross-encoder based on microsoft/deberta-v3-base. It was trained on Ms-Marco using loss infoNCE 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 infoNCE

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-infoNCE")

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	38.97	45.66
trec2019	96.51	73.66
trec2020	93.58	71.28
fever	74.33	74.45
arguana	15.10	22.46
climate_fever	17.21	12.95
dbpedia	69.25	40.12
fiqa	45.74	37.93
hotpotqa	82.35	65.70
nfcorpus	34.44	19.40
nq	51.79	56.80
quora	63.02	66.26
scidocs	27.05	15.40
scifact	64.81	67.24
touche	63.53	35.51
trec_covid	91.92	76.07
robust04	66.19	43.56
lotte_writing	69.59	60.22
lotte_recreation	62.20	57.07
lotte_science	48.70	40.64
lotte_technology	56.84	47.53
lotte_lifestyle	72.84	64.26
Mean In Domain	76.35	63.53
BEIR 13	53.89	45.41
LoTTE (OOD)	62.73	52.21