OpenReview
.net
OpenReview
.net
Login
OpenReview
.net
Login
Diego Marco Granziol
Researcher, University of Oxford
Joined
January 2025
Names
Diego Marco Granziol
(Preferred)
,
Diego Granziol
Emails
****@robots.ox.ac.uk
(Confirmed)
,
****@purestrength.ai
(Confirmed)
,
****@maths.ox.ac.uk
(Confirmed)
Personal Links
Google Scholar
ORCID
Career & Education History
Researcher
University of Oxford
(oxford.ac.uk)
2022
–
Present
PhD student
Engineering,
University of Oxford
(oxford.ac.uk)
2016
–
2020
Advisors, Relations & Conflicts
Coauthor
Nicholas Baskerville
Present
Coauthor
Xingchen Wan
Present
PhD Advisor
Michael Osborne
2017
–
2020
PhD Advisor
S Roberts
2016
–
2020
Expertise
Social Network Analysis
2018
–
Present
Information Theory
2016
–
Present
Statistical Mechanics
2016
–
Present
Deep Learning
2016
–
2020
Publications
Compute-Optimal LLMs Provably Generalize Better with Scale
Marc Anton Finzi
,
Sanyam Kapoor
,
Diego Granziol
,
Anming Gu
,
Christopher De Sa
,
J Zico Kolter
,
Andrew Gordon Wilson
ICLR 2025 Poster
Readers:
Everyone
A Practical PAC-Bayes Generalisation Bound for Deep Learning
Diego Granziol
,
Mingtian Zhang
,
Nicholas Baskerville
29 Sept 2021 (modified: 13 Feb 2023)
ICLR 2022 Conference Withdrawn Submission
Readers:
Everyone
Deep Curvature Suite
Diego Granziol
,
Xingchen Wan
,
Timur Garipov
28 Sept 2020 (modified: 05 May 2023)
ICLR 2021 Conference Blind Submission
Readers:
Everyone
Flatness is a False Friend
Diego Granziol
28 Sept 2020 (modified: 05 May 2023)
ICLR 2021 Conference Blind Submission
Readers:
Everyone
Fast Information-theoretic Bayesian Optimisation
Bin Xin Ru
,
Mark McLeod
,
Diego Granziol
,
Michael A. Osborne
2018 (modified: 11 Nov 2022)
ICML 2018
Readers:
Everyone
Co-Authors
Andrew Gordon Wilson
Anming Gu
Bin Xin Ru
Christopher De Sa
J Zico Kolter
Marc Anton Finzi
Mark McLeod
Michael A. Osborne
Mingtian Zhang
Nicholas Baskerville
Sanyam Kapoor
Timur Garipov
Xingchen Wan