33.2430, FYI: Inverse Scaling Prize ($100k grand prize) submissions are open!

[The LINGUIST List]

LINGUIST List: Vol-33-2430. Sat Aug 06 2022. ISSN: 1069 - 4875.

Subject: 33.2430, FYI: Inverse Scaling Prize ($100k grand prize) submissions are open!

Moderator: Malgorzata E. Cavar (linguist at linguistlist.org)
Student Moderator: Billy Dickson
Managing Editor: Lauren Perkins
Team: Helen Aristar-Dry, Everett Green, Sarah Goldfinch, Nils Hjortnaes,
        Joshua Sims, Billy Dickson, Amalia Robinson, Matthew Fort
Jobs: jobs at linguistlist.org | Conferences: callconf at linguistlist.org | Pubs: pubs at linguistlist.org

Homepage: http://linguistlist.org

Hosted by Indiana University

Please support the LL editors and operation with a donation at:
           https://funddrive.linguistlist.org/donate/

Editor for this issue: Everett Green <everett at linguistlist.org>
================================================================


Date: Sat, 06 Aug 2022 08:44:23
From: Najoung Kim [najoung.kim at nyu.edu]
Subject: Inverse Scaling Prize ($100k grand prize) submissions are open!

 
NYU is announcing the Inverse Scaling Prize: a $100k grand prize + $150k in
additional prizes for finding an important task where larger language models
do worse.

We’re running submissions in two rounds, with the first deadline on August 27.
Details here: https://github.com/inverse-scaling/prize . We want to emphasize
that you don’t need extensive experience with machine learning to participate,
and we are very interested in submissions from researchers studying language
outside of machine learning, including theoretical and experimental
linguistics. 

Motivation & background: Pretrained language models (LMs)—neural network
models trained on the objective of context reconstruction (e.g., predicting
the next token in a sequence or predicting tokens from masked out
contexts)---have contributed to rapid progress in natural language processing.
It has also been suggested that larger models (in terms of model size and
training data) consistently, predictably do better than smaller ones on many
tasks (“scaling laws”). However, scaling doesn’t always improve LMs on all
axes. For instance, social biases & toxicity have been shown to get amplified.
This contest is a call for important tasks where models actively get worse
with scale (i.e., “inverse scaling”).

Such tasks seem rare, but we’ve found some. E.g., in one question answering
task, we’ve noticed that asking a question while including your beliefs
influences larger models more than smaller models to be biased towards your
belief. Other possible examples are imitating mistakes/bugs in the prompt or
repeating common misconceptions.

Finding more examples of inverse scaling would point to important issues with
using large language models that won’t go away with scale. These examples
could provide inspiration for improving our models in the future, for instance
by designing better datasets and/or training objectives.

If it turns out to be very difficult to find inverse scaling, that would be
some evidence that LM scaling would not make LMs worse in noticeable ways in
the near term.

To enter the contest:
1) Identify a task that you suspect shows inverse scaling
2) Construct a dataset of 300+ examples for the task
3) Test your dataset for inverse scaling with GPT-3/OPT using our Colab
notebooks
4) Follow instructions here to submit:
https://github.com/inverse-scaling/prize

Submissions will be evaluated on a series of private models provided by
Anthropic, and prize decisions will be made by a panel of anonymous reviewers.

For questions, reach out to us at inverse.scaling [at] gmail.com, open an
issue on our repository, or join our Slack (details in our repository:
https://github.com/inverse-scaling/prize).

We’re excited for people from all fields to take part (philosophy, cog sci,
linguistics, etc), and we’ve designed our tools to be easy for machine
learning newcomers to use too.

The Inverse Scaling Prize is run by a team out of New York University: Ian
McKenzie, Alex Lyzhov, Alicia Parrish, Ameya Prabhu, Aaron Mueller, Najoung
Kim, Sam Bowman, and Ethan Perez. The prize pool was generously made available
by Future Fund.

If you’re excited about the contest, we’d appreciate you sharing it with
people who might be interested in participating.
 



Linguistic Field(s): Computational Linguistics





 



------------------------------------------------------------------------------

***************************    LINGUIST List Support    ***************************
 The 2020 Fund Drive is under way! Please visit https://funddrive.linguistlist.org
  to find out how to donate and check how your university, country or discipline
     ranks in the fund drive challenges. Or go directly to the donation site:
                   https://crowdfunding.iu.edu/the-linguist-list

                        Let's make this a short fund drive!
                Please feel free to share the link to our campaign:
                    https://funddrive.linguistlist.org/donate/
 


----------------------------------------------------------
LINGUIST List: Vol-33-2430	
----------------------------------------------------------