ranxhub

Run

RanxHub ID: msmarco-passage/dev/ranxhub/ance
Version: 1.0
Description: ANCE run reproduced using Pyserini.
Tags: Retrieval
Date: 3 February 2023
Run Authors: Elias Bassani
From Paper:
Paper Authors:

Model

Name: ANCE
Description: Single representation, dense retrieval model trained by selecting global negatives with big gradient norms using an asynchronously updated ANN index.
Tags: Single Representation · Dense Retrieval · Dot Product · BERT
Paper: Approximate nearest neighbor negative contrastive learning for dense text retrieval
Authors: Lee Xiong · Chenyan Xiong · Ye Li · Kwok-Fung Tang · Jialin Liu · Paul Bennett · Junaid Ahmed · Arnold Overwijk

Results

MRR@10 MAP@1000 Recall@1000
0.33014502888070213 0.33623895943583004 0.958655682903534