cross-encoder-ettin-68m-BCE

This model is a cross-encoder based on jhu-clsp/ettin-encoder-68m. It was trained on Ms-Marco using loss bce 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 bce

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-BCE")
model = AutoModelForSequenceClassification.from_pretrained("xpmir/cross-encoder-ettin-68m-BCE")

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	34.09	40.72
trec2019	88.98	65.52
trec2020	86.88	64.25
fever	74.36	74.93
arguana	13.58	19.34
climate_fever	13.37	9.62
dbpedia	62.96	34.72
fiqa	42.67	34.21
hotpotqa	80.90	63.82
nfcorpus	43.92	24.92
nq	46.00	51.23
quora	71.33	73.68
scidocs	22.35	12.45
scifact	57.45	59.11
touche	53.79	29.25
trec_covid	90.29	67.40
robust04	46.58	27.32
lotte_writing	66.33	57.78
lotte_recreation	57.76	52.61
lotte_science	41.95	36.62
lotte_technology	49.58	41.97
lotte_lifestyle	66.95	58.78
Mean In Domain	69.98	56.83
BEIR 13	51.77	42.67
LoTTE (OOD)	54.86	45.85