cross-encoder-ELECTRA-Hinge

This model is a cross-encoder based on google/electra-base-discriminator. 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.

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 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("google/electra-base-discriminator")
model = AutoModelForSequenceClassification.from_pretrained("xpmir/cross-encoder-ELECTRA-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	39.19	45.79
trec2019	95.23	72.98
trec2020	95.06	73.29
fever	78.60	78.65
arguana	17.81	26.69
climate_fever	25.29	18.99
dbpedia	74.04	44.10
fiqa	48.50	40.12
hotpotqa	87.76	70.32
nfcorpus	56.74	34.24
nq	52.83	57.95
quora	77.72	79.87
scidocs	27.42	15.71
scifact	64.86	67.55
touche	66.31	36.31
trec_covid	91.22	68.55
robust04	70.82	46.89
lotte_writing	70.50	61.09
lotte_recreation	62.30	56.98
lotte_science	48.48	39.77
lotte_technology	56.42	47.25
lotte_lifestyle	74.14	64.54
Mean In Domain	76.49	64.02
BEIR 13	59.16	49.16
LoTTE (OOD)	63.78	52.75