squad-extraction

This is a project page for our ICLR 2020 paper on model extraction attacks on BERT. For more details, contact me at kalpesh@cs.umass.edu.

Thieves on Sesame Street! Model Extraction of BERT-based APIs
Kalpesh Krishna, Gaurav Singh Tomar, Ankur P. Parikh, Nicolas Papernot, Mohit Iyyer

arXiv: https://arxiv.org/abs/1910.12366
OpenReview: https://openreview.net/forum?id=Byl5NREFDr

blogpost: http://www.cleverhans.io/2020/04/06/stealing-bert.html
Twitter: https://twitter.com/kalpeshk2011/status/1247550470454984705

codebase: https://github.com/google-research/language/tree/master/language/bert_extraction
slide deck: https://docs.google.com/presentation/d/1emJb07ER8Gje6PGauIU–t0DCgvECJlnjVQfv9rOUkM/edit?usp=sharing
ICLR presentation & chat: https://iclr.cc/virtual_2020/poster_Byl5NREFDr.html

external coverage: YouTube summary by TechViz // NLPHighlights Podcast // Sebastian Ruder’s NLP News // HubofML Newsletter // TopBot’s Best NLP Papers at ICLR 2020 // VentureBeat ICLR 2020 Cover Photo // MadeWithML