cross-encoder-MiniLM-L12-DistillRankNET

This model is a cross-encoder based on microsoft/MiniLM-L12-H384-uncased. 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("microsoft/MiniLM-L12-H384-uncased")
model = AutoModelForSequenceClassification.from_pretrained("xpmir/cross-encoder-MiniLM-L12-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	37.40	43.98
trec2019	96.12	74.57
trec2020	93.83	73.48
fever	81.21	80.95
arguana	18.48	27.97
climate_fever	27.52	20.31
dbpedia	75.81	46.06
fiqa	43.71	36.25
hotpotqa	85.35	66.48
nfcorpus	57.75	34.59
nq	53.19	58.21
quora	76.34	78.62
scidocs	28.06	15.79
scifact	66.12	69.34
touche	64.33	34.46
trec_covid	87.17	70.74
robust04	75.25	52.28
lotte_writing	66.66	58.11
lotte_recreation	60.60	55.12
lotte_science	46.01	38.34
lotte_technology	53.36	44.41
lotte_lifestyle	71.62	61.69
Mean In Domain	75.78	64.01
BEIR 13	58.85	49.21
LoTTE (OOD)	62.25	51.66