Skip to main content
Cornell University
We gratefully acknowledge support from the Simons Foundation, member institutions, and all contributors. Donate
arxiv logo > cs > arXiv:2203.03466

Help | Advanced Search

arXiv logo
Cornell University Logo

quick links

  • Login
  • Help Pages
  • About

Computer Science > Machine Learning

arXiv:2203.03466 (cs)
[Submitted on 7 Mar 2022 (v1), last revised 28 Mar 2022 (this version, v2)]

Title:Tensor Programs V: Tuning Large Neural Networks via Zero-Shot Hyperparameter Transfer

Authors:Greg Yang, Edward J. Hu, Igor Babuschkin, Szymon Sidor, Xiaodong Liu, David Farhi, Nick Ryder, Jakub Pachocki, Weizhu Chen, Jianfeng Gao
View a PDF of the paper titled Tensor Programs V: Tuning Large Neural Networks via Zero-Shot Hyperparameter Transfer, by Greg Yang and 9 other authors
View PDF
Abstract:Hyperparameter (HP) tuning in deep learning is an expensive process, prohibitively so for neural networks (NNs) with billions of parameters. We show that, in the recently discovered Maximal Update Parametrization (muP), many optimal HPs remain stable even as model size changes. This leads to a new HP tuning paradigm we call muTransfer: parametrize the target model in muP, tune the HP indirectly on a smaller model, and zero-shot transfer them to the full-sized model, i.e., without directly tuning the latter at all. We verify muTransfer on Transformer and ResNet. For example, 1) by transferring pretraining HPs from a model of 13M parameters, we outperform published numbers of BERT-large (350M parameters), with a total tuning cost equivalent to pretraining BERT-large once; 2) by transferring from 40M parameters, we outperform published numbers of the 6.7B GPT-3 model, with tuning cost only 7% of total pretraining cost. A Pytorch implementation of our technique can be found at this http URL and installable via `pip install mup`.
Comments: NeurIPS 2021
Subjects: Machine Learning (cs.LG); Disordered Systems and Neural Networks (cond-mat.dis-nn); Neural and Evolutionary Computing (cs.NE)
Cite as: arXiv:2203.03466 [cs.LG]
  (or arXiv:2203.03466v2 [cs.LG] for this version)
  https://doi.org/10.48550/arXiv.2203.03466
arXiv-issued DOI via DataCite

Submission history

From: Greg Yang [view email]
[v1] Mon, 7 Mar 2022 15:37:35 UTC (2,817 KB)
[v2] Mon, 28 Mar 2022 08:12:14 UTC (2,818 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled Tensor Programs V: Tuning Large Neural Networks via Zero-Shot Hyperparameter Transfer, by Greg Yang and 9 other authors
  • View PDF
  • TeX Source
  • Other Formats
view license
Current browse context:
cs.LG
< prev   |   next >
new | recent | 2022-03
Change to browse by:
cond-mat
cond-mat.dis-nn
cs
cs.NE

References & Citations

  • NASA ADS
  • Google Scholar
  • Semantic Scholar

2 blog links

(what is this?)
a export BibTeX citation Loading...

BibTeX formatted citation

×
Data provided by:

Bookmark

BibSonomy logo Reddit logo

Bibliographic and Citation Tools

Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)

Code, Data and Media Associated with this Article

alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)

Demos

Replicate (What is Replicate?)
Hugging Face Spaces (What is Spaces?)
TXYZ.AI (What is TXYZ.AI?)

Recommenders and Search Tools

Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
IArxiv Recommender (What is IArxiv?)
  • Author
  • Venue
  • Institution
  • Topic

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)
  • About
  • Help
  • contact arXivClick here to contact arXiv Contact
  • subscribe to arXiv mailingsClick here to subscribe Subscribe
  • Copyright
  • Privacy Policy
  • Web Accessibility Assistance
  • arXiv Operational Status
    Get status notifications via email or slack