cross-encoder-ettin-32m-ADR-MSE

This model is a cross-encoder based on jhu-clsp/ettin-encoder-32m. 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-32m-ADR-MSE")
model = AutoModelForSequenceClassification.from_pretrained("xpmir/cross-encoder-ettin-32m-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	30.21	35.72
trec2019	83.41	62.03
trec2020	85.55	62.79
fever	71.84	72.44
arguana	7.77	11.69
climate_fever	21.49	15.47
dbpedia	63.40	34.50
fiqa	33.97	25.77
hotpotqa	68.51	50.93
nfcorpus	44.83	25.25
nq	37.55	42.18
quora	71.95	73.20
scidocs	17.86	9.43
scifact	48.13	50.36
touche	66.33	31.05
trec_covid	80.14	60.85
robust04	57.89	33.16
lotte_writing	60.22	50.65
lotte_recreation	49.07	44.49
lotte_science	40.88	34.02
lotte_technology	41.87	34.09
lotte_lifestyle	59.49	50.96
Mean In Domain	66.39	53.51
BEIR 13	48.75	38.70
LoTTE (OOD)	51.57	41.23