cross-encoder-ettin-17m-ADR-MSE

This model is a cross-encoder based on jhu-clsp/ettin-encoder-17m. It was trained on Ms-Marco using loss ADR 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 ADR

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-17m-ADR-MSE")
model = AutoModelForSequenceClassification.from_pretrained("xpmir/cross-encoder-ettin-17m-ADR-MSE")

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	21.11	25.77
trec2019	74.68	49.64
trec2020	79.30	49.10
fever	61.48	62.98
arguana	7.58	11.92
climate_fever	13.31	9.82
dbpedia	47.05	23.87
fiqa	23.28	17.70
hotpotqa	57.88	41.76
nfcorpus	36.09	20.18
nq	27.06	30.55
quora	66.70	67.76
scidocs	13.01	6.88
scifact	46.78	50.01
touche	51.27	26.50
trec_covid	84.62	53.35
robust04	44.94	23.63
lotte_writing	45.64	36.52
lotte_recreation	36.77	32.87
lotte_science	28.17	22.75
lotte_technology	28.72	22.60
lotte_lifestyle	47.76	39.53
Mean In Domain	58.36	41.50
BEIR 13	41.24	32.56
LoTTE (OOD)	38.67	29.65