Leshem Choshen · @LChoshen
770 followers · 146 posts · Server sigmoid.social

RT @LChoshen@twitter.com

🧞‍♂️Human annotations for free🥳

As each one reinvents their annotation process GENIE🧞
Just decided to standardize the whole process (API MTurk...)
Upload your process and
the next papers would be able to exactly replicate
And they pay for annotation!

🐦🔗: twitter.com/LChoshen/status/16

#emnlp2022livetweet

Last updated 3 years ago

Leshem Choshen · @LChoshen
770 followers · 146 posts · Server sigmoid.social

RT @d_aumiller@twitter.com

And the final from the morning: Aviv Slobodkin explaining the general idea behind their new task of what they call "controlled text reduction".
Interesting similarities to QA, but also exciting avenue to take summarization research to the next level.

🐦🔗: twitter.com/d_aumiller/status/

#emnlp2022livetweet

Last updated 3 years ago

Leshem Choshen · @LChoshen
762 followers · 141 posts · Server sigmoid.social

Simplification for cognitive disabilities

Simplifying for specific disabilities, not to undefined readers

they propose 1⃣st
Dataset
Needed operations taxonomy
Model

@Eytan_Chamovitz@twitter.com @AbendOmri@twitter.com @nlphuji@twitter.com
(in delay)

#emnlp2022livetweet

Last updated 3 years ago

Leshem Choshen · @LChoshen
762 followers · 139 posts · Server sigmoid.social

🧞‍♂️Human annotations for free🥳

As each one reinvents their annotation process GENIE🧞
Just decided to standardize the whole process (API MTurk...)
Upload your process and
the next papers would be able to exactly replicate
And they pay for annotation!

#emnlp2022livetweet

Last updated 3 years ago

Leshem Choshen · @LChoshen
762 followers · 132 posts · Server sigmoid.social

Efficient finetuning can predict transferrability
Tuning changes some weights
If different weights are changed similarly -> similar tasks
Indeed this predicts inter-training improvements better than previous works
link👇
@wangchunshu@twitter.com @JulianMcauley@twitter.com @XuCanwen@twitter.com

#emnlp2022livetweet

Last updated 3 years ago

Leshem Choshen · @LChoshen
761 followers · 125 posts · Server sigmoid.social

Comics is a language (keynote)🧵

language has many modalities that often mix
(fig text+images but also speech, sign...)
let's understand the language of paintings

@neilcohn

#emnlp2022livetweet

Last updated 3 years ago

Leshem Choshen · @LChoshen
761 followers · 125 posts · Server sigmoid.social

Comics is a language (keynote)🧵

language has many modalities that often mix
(fig text+images but also speech, sign...)
let's understand the language of paintings

@visual_linguist@twitter.com

#emnlp2022livetweet

Last updated 3 years ago

Boyang "Albert" Li · @AlbertBoyangLi
23 followers · 15 posts · Server sigmoid.social

Second day at . My personal favorite of the day: Razvan Pascanu's keynote at the Multilingual Representation Learning Workshop. Some ideas here will inform the next decade, I believe.

#emnlp2022 #emnlp2022livetweet

Last updated 3 years ago

Leshem Choshen · @LChoshen
759 followers · 121 posts · Server sigmoid.social

Methapor dataset with some twists
They annotated Spanish 117K tokens (3.6K sentences) for being metaphoric or not

🟦new dataset
🟦2-10% of sentences have a metaphor🤨
🟦Cross lingual - SoTA on Spanish, and they didn't even multitask


linkes👇

#emnlp2022livetweet #conll2022

Last updated 3 years ago

Leshem Choshen · @LChoshen
759 followers · 120 posts · Server sigmoid.social

Tydip: a dataset of politeness in 9 languages

Reasonable agreement and classification scores based on it

Anirudh Srinivasan @eunsolc@twitter.com
arxiv.org/abs/2211.16496
underline.io/events/342/sessio

#emnlp2022livetweet #conll2022

Last updated 3 years ago

Leshem Choshen · @LChoshen
759 followers · 119 posts · Server sigmoid.social

The old man the ship? Easy

LMs underestimate the surprisal of garden path sentences and syntax
Controlling for words complexity (lexical)

@sArehalli@twitter.com @linguistbrian@twitter.com @tallinzen@twitter.com
arxiv.org/abs/2210.12187
underline.io/events/342/poster

#emnlp2022livetweet #conll2022

Last updated 3 years ago

Leshem Choshen · @LChoshen
759 followers · 118 posts · Server sigmoid.social

Mismatches between LMs and humans

Jhon/Johanna knew Robert\a hated *him*

Humans&GPTs find "him" surprising after Johana
But GPTs are also surprised when they shouldn't be

@Forrest_L_Davis@twitter.com
(No arXiv?)
underline.io/events/342/poster

#emnlp2022livetweet #conll2022

Last updated 3 years ago

Leshem Choshen · @LChoshen
759 followers · 117 posts · Server sigmoid.social

Semantics is not only BERT
DALLE-2:
Instead of disambiguating uses information from all meanings of the words simultaneously

(cool pictures that we don't usually have as a bonus)
arxiv.org/abs/2210.10606
@RoyiRassin@twitter.com @ravfogel@twitter.com @yoavgo@twitter.com

#nlproc #emnlp2022livetweet #conll2022

Last updated 3 years ago

Leshem Choshen · @LChoshen
757 followers · 113 posts · Server sigmoid.social

I will be at @emnlpmeeting@twitter.com & @conll_conf@twitter.com say hi

I will be tweeting under or or or

If it spams you, mute it (or wait a week 😉)
help.twitter.com/en/using-twit

#emnlp2022livetweet #emnlp2022 #conll #conlllivetweet2022

Last updated 3 years ago