Jurek Leonhardt
@mr_jleo
Followers
49
Following
25
Media
0
Statuses
12
Joined April 2022
🚀 Our #TOIS journal paper showcases the power of GPU-free text ranking using lightweight encoders and forward indexes. This is @mr_jleo's for his final thesis chapter. Kudos to co-authors Henrik Müller, @krudra5, @abhijit_ai, @meghakhosla
dl.acm.org
Dual-encoder-based dense retrieval models have become the standard in IR. They employ large Transformer-based language models, which are notoriously inefficient in terms of resources and latency. We...
0
3
23
2nd #TOIS paper is on Data Augmentation for Sample Efficient & Robust Document Ranking. We show performance boosts across various dataset sizes, w/ impressive out-of-domain transferability. Kudos to lead @abhijit_ai and @mr_jleo, @krudra5, @movedthecheese 🚀 paper coming soon.
0
1
16
📢SEA in October - Efficiency in neural IR 🗓️ Oct 27th, 17:00 CET, in-person at Lab42 or online✍️ Sing up now! Speakers: -Carlos Lassance (Naverlabs) @cadurosar -Jurek Leonhardt (TU Delft) @mr_jleo More info, and registration link: https://t.co/RE5vdSH0w1
@irlab_amsterdam
meetup.com
In this edition of SEA we will discuss efficiency in neural IR. We have two amazing speakers lined up: Jurek Leonhardt (TU Delft) and Carlos Lassance (Naverlabs). This will
0
4
11
Very happy that this got through with @mr_jleo . In this work we propose how to dynamically select extractive document text as explanations for document ranking. Turns out that ranking just on these explanations does not negotiate ranking performance. #tois
"Extractive Explanations for Interpretable Text Ranking" by Jurek Leonhardt,Koustav Rudra,Avishek Anand accepted in ACM Transactions on Information Systems (Just Accepted)
2
4
23
Happy to share our survey about explainable IR, comments are appreciated. @run4avi @maxidahl @JonasWallat @YumengWang13 @Joshua_Ghost
Explainable Information Retrieval: A Survey 🔗:
0
11
23
#xaiss ended with awesome hackathon presentations and prizes🏅🏆 Privileged to have such an awesome team @mr_jleo @andradenisio @alisarieg @LijunLyu @timdr4ws @_Guz_
1
4
15
And we are underway..#xaiss starts with the Keynote from @MihaelaVDS on New frontiers in ML interpretability.
0
8
24
📢A few days left to register for the Explainable AI Summer School📢 We've prepared a series of hot 🔥engaging 📖topics, and fun activities!🥳 Discover XAI, expand you knowledge, and explore beautiful Delft 🇳🇱 🚨 Checkout the schedule and register on
0
3
7
Don't forget to register for @SoBigData summer school on #eXplainableAI at @tudelft
https://t.co/wiN1l6QtM8
0
3
6
We are organising a summer school on Explainable AI in beautiful Delft from Aug. 29 - Sept 2. Great lectures, Hands on material, Hackathon and Amazing socials. Registration open..spots limited. https://t.co/hfcxfCOWtc
#sobigdata #l3s #tudelft #xaiss
1
18
47
What if you wanted to retrieve and rank documents < 200ms using CPUs and still be competitive to BERT-based ranking? Meet Fast-Forward indexes paper: https://t.co/ClkNKpBTkK slides: https://t.co/tnRqYVJzJB repo: https://t.co/QbFzEveWey
#TheWebConference22
arxiv.org
Neural document ranking approaches, specifically transformer models, have achieved impressive gains in ranking performance. However, query processing using such over-parameterized models is both...
1
7
32