cross-encoder-ettin-150m-infoNCE

This model is a cross-encoder based on jhu-clsp/ettin-encoder-150m. 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-150m-infoNCE")
model = AutoModelForSequenceClassification.from_pretrained("xpmir/cross-encoder-ettin-150m-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	41.18	47.80
trec2019	96.40	75.27
trec2020	93.83	73.24
fever	81.69	81.36
arguana	18.21	27.48
climate_fever	30.59	22.75
dbpedia	76.20	46.31
fiqa	50.71	41.99
hotpotqa	89.99	73.59
nfcorpus	56.47	36.13
nq	54.85	59.89
quora	76.53	79.03
scidocs	29.55	17.22
scifact	57.77	63.58
touche	63.36	36.89
trec_covid	94.50	78.88
robust04	66.74	47.46
lotte_writing	77.07	68.14
lotte_recreation	65.45	60.32
lotte_science	52.17	43.57
lotte_technology	60.63	51.70
lotte_lifestyle	75.67	66.26
Mean In Domain	77.14	65.44
BEIR 13	60.03	51.16
LoTTE (OOD)	66.29	56.24