cross-encoder-ettin-68m-infoNCE

This model is a cross-encoder based on jhu-clsp/ettin-encoder-68m. 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("xpmir/cross-encoder-ettin-68m-infoNCE")
model = AutoModelForSequenceClassification.from_pretrained("xpmir/cross-encoder-ettin-68m-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	40.21	46.90
trec2019	95.04	73.74
trec2020	94.75	72.76
fever	77.53	77.61
arguana	15.59	22.32
climate_fever	22.02	16.08
dbpedia	72.45	43.43
fiqa	47.94	39.50
hotpotqa	86.41	70.27
nfcorpus	53.89	33.28
nq	52.46	57.73
quora	78.53	80.60
scidocs	28.38	15.73
scifact	67.40	70.03
touche	63.39	36.10
trec_covid	89.83	75.08
robust04	67.90	45.38
lotte_writing	72.76	64.23
lotte_recreation	63.21	57.33
lotte_science	52.26	42.76
lotte_technology	56.39	48.19
lotte_lifestyle	72.71	63.47
Mean In Domain	76.67	64.47
BEIR 13	58.14	49.06
LoTTE (OOD)	64.21	53.56