cross-encoder-ettin-68m-DistillRankNET

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

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

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	34.13	40.46
trec2019	98.84	74.07
trec2020	91.51	71.92
fever	74.80	75.23
arguana	14.35	21.10
climate_fever	16.46	12.19
dbpedia	71.85	42.73
fiqa	41.85	34.25
hotpotqa	84.44	66.11
nfcorpus	53.60	32.11
nq	48.74	53.78
quora	76.33	78.23
scidocs	23.82	12.96
scifact	58.10	60.19
touche	60.96	35.49
trec_covid	91.02	75.07
robust04	66.32	42.11
lotte_writing	70.84	61.51
lotte_recreation	57.85	52.52
lotte_science	48.94	40.54
lotte_technology	52.33	43.25
lotte_lifestyle	69.70	60.54
Mean In Domain	74.83	62.15
BEIR 13	55.10	46.11
LoTTE (OOD)	61.00	50.08