cross-encoder-ELECTRA-BCE

This model is a cross-encoder based on google/electra-base-discriminator. 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("google/electra-base-discriminator")
model = AutoModelForSequenceClassification.from_pretrained("xpmir/cross-encoder-ELECTRA-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	38.30	44.84
trec2019	96.26	68.76
trec2020	93.94	68.94
fever	77.33	77.78
arguana	15.62	23.70
climate_fever	26.30	19.79
dbpedia	73.63	42.74
fiqa	45.47	37.62
hotpotqa	85.70	68.92
nfcorpus	41.10	23.54
nq	51.94	57.27
quora	67.47	71.44
scidocs	26.00	14.61
scifact	62.56	64.84
touche	57.63	32.82
trec_covid	78.92	53.58
robust04	59.27	39.99
lotte_writing	67.76	58.85
lotte_recreation	60.53	55.99
lotte_science	43.00	35.64
lotte_technology	54.21	44.85
lotte_lifestyle	72.41	62.84
Mean In Domain	76.17	60.85
BEIR 13	54.59	45.28
LoTTE (OOD)	59.53	49.69