Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Jacob Andreas Profile picture
@jacobandreas
Jul 31, 2017 • 5 tweets • 1 min read • Read on X
Next: Ravi Shekhar on breaking captioning models arxiv.org/abs/1705.01359 #acl2017nlp
Take a captioned image, replace one noun with a different noun that makes it a false caption. Can do this automatically w/ COCO cat labels

• • •

Missing some Tweet in this thread? You can try to force a refresh
　

Keep Current with Jacob Andreas

Jacob Andreas Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @jacobandreas

Jacob Andreas Profile picture
Dec 9, 2022
I wish the research community had good shared shorthand for "This ML model can X {syntax, meaning, intent, ...}", where X is roughly "represent and use, with less robustness and systematicity than humans but greater success than any trivial baseline we know of".
Much of the current "can LMs model meaning" discourse is disagreement not about "meaning", but "model". There are lots of interesting intermediate states between total failure and total success. "Can model meaning" suggests a binary criterion & we all have different thresholds.
In particular, I (and I think many others who grew up in the statistical NLP tradition) still think of n-gram models as "prototypical" language generation systems. Now we wake up every day and say "holy crap, a talking n-gram model" (mit.edu/people/dpolica…)
Read 5 tweets
Jacob Andreas Profile picture
Dec 6, 2022
Speculative (!!!) paper arguing that big LMs can model agency & communicative intent: arxiv.org/abs/2212.01681 (somehow in EMNLP findings). Briefly:

1. LMs do not in general have beliefs or goals. An LM trained on the Internet models a distribution over next tokens *marginalized* First page of the paper "Language Models as Agent Model
over all the authors who could have produced the context. All large-scale training sets are generated by a mixture of authors w/ mutually incompatible beliefs & goals, and we shouldn't expect a model of the marginal dist. over utts to be coherent.

BUT
2. B/c single training documents are written by individuals who do have specific communicative intentions, and b/c understanding these intentions helps w/ next-word prediction, we should expect LMs to *infer and represent* the latent beliefs/intentions/etc that give rise to a ctx
Read 8 tweets
Jacob Andreas Profile picture
Sep 21, 2020
New TACL paper from Semantic Machines! A first look at the approach we've been developing for modeling complex, task-oriented conversations using dataflow graphs.

aka.ms/AA9oxf3

I am extremely excited about this work for several reasons:
1/ It highlights one of the most consequential but overlooked consequences of neural models in dialogue: explicit representations of intent still matter (a production system can't lie about what it did even 1% of the time) but now *we can predict whatever representations we want*
If you look at the earliest dialogue research (e.g. Grosz & Sidner's Shared Plans), the community used to be way more ambitious about the dialogue phenomena we tried to represent. Most of that went out the window when contemporary ML approaches weren't up to modeling it.
Read 9 tweets
Jacob Andreas Profile picture
Jul 9, 2020
Was also very happy to get multiple pointers to @alexandersclark & Eyraud's work on substitutable languages (dl.acm.org/doi/pdf/10.555…). Haven't done a full read yet but "weak substitutability <=> syntactic congruence" is exactly what (1-fragment, full-context) GECA assumes---
so you can think of GECA as attempting to constrain models to substitutable languages by computing the closure of the training data over observed syntactic congruences. This closure need not produce a CFL (which answers a question I had about GECA!)...
...and I think the constraint could be straightforwardly extended to *unobserved* sequences using a posterior regularization approach like this one from @XiangLisaLi2 and @srush_nlp virtual.acl2020.org/paper_main.243….
Read 5 tweets
Jacob Andreas Profile picture
Jul 6, 2020
Thoughts on Kathy McKeown's #acl2020nlp keynote:
someone should go back and categorize all the "what's most important about deep learning" responses---literally nobody agrees! Accessibility, scalability, empirical performance, representations of lexical meaning, feature representations more broadly, support for new inputs, ...
Always good to be reminded that apart from Aravind, almost all of the "founding members" of what's now the NLP community were women: Spärck-Jones, Webber, Grosz, Hajicova.
Read 7 tweets
Jacob Andreas Profile picture
Jul 3, 2020
sing minutes grammars and three transducers
neural semantic persons
invariant to lexical choice by replacing words with their citizens
Read 8 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

Send Email!

Become Premium to read this thread!

Read all the old archives from Twitter by becoming our Premium member for only $3/month! Read more »

Become Premium

Already a Premium member? Login to access this thread

:(