Johnny Ma

Content Data Scientist at HBOMax. Hungry for fries.

[David Bamman @dbamman]

Happy to get this EMNLP paper by @_akpiper @RichardJeanSo and myself out there: "Narrative Theory for Computational Narrative Understanding" https://t.co/i2onKfwOvf

[Johnny Ma @_JohnnyMa_]

@Ted_Underwood Yes, almost for sure. It’s actually a problem when building new datasets for long form QA or narrative summarization because you can’t include any popular media that might have a wikipedia, goodreads, or reddit page that a pretrained model would have already consumed.

[Johnny Ma @_JohnnyMa_]

Seems like three types of content consumers: the Explorer, the Fanatic, and the Sampler.

[Netflix @netflix]

PLAY SOMETHING will show you:

✔️ A new series/film similar to one you've watched before

✔️An episode/film you’ve already watched and may want to watch again if it's been a while

✔️ An episode from a show you've started but haven't finished (it will pick up where you left off)

[Johnny Ma @_JohnnyMa_]

@peligrietzer @beausievers BART is still trained on predicting masked spans though, not reconstructing the whole input. Though it occasionally does if the span is sentence size.

[Johnny Ma @_JohnnyMa_]

For our school's day off, I spent some time learning d3.js and css to make an interactive trainee ranking graphic for #iQiyi's #YouthWithYou3. I will update it for each new ranking. Enjoy!

https://t.co/pUxl6FWUjz

#青春有你3 #QCYN3 #ywy3 #爱奇艺

[Johnny Ma @_JohnnyMa_]

@_akpiper take stories and remove elements until they are no longer considered stories? that way you can determine the relative breaking point

[Johnny Ma @_JohnnyMa_]

@NytReview ahhh the plural, thanks. rough one.

[Johnny Ma @_JohnnyMa_]

@NytReview could you explain what’s going on? still don’t get it...

[Johnny Ma @_JohnnyMa_]

@WaterWish298 no idea either

[Johnny Ma @_JohnnyMa_]

@alexisjross Love the work! Is the code available anywhere ?

[Johnny Ma @_JohnnyMa_]

looks great, a bold bet on the future. but nothing beats the real thing, cannot wait to go back.

[Warner Bros. @warnerbros]

I got you something ✨nice✨ this year:
🎁 The biggest movie premieres
🎁 In theaters and on HBO Max the exact same day
🎁 Beginning December 25 with #WonderWoman1984
#HBOMax #WBPictures
https://t.co/QA8MlErRYQ

[Johnny Ma @_JohnnyMa_]

absolutely will do this for manga panels this winter break

[Emil Wallner @EmilWallner]

*** It’s LIVE ***

I've made an index with 340 creative Machine Learning projects!

More coming every week!!

------------------------
👉 https://t.co/P7OdrKAk1X 👈
------------------------

(*・ω・)ﾉ🍾 YAY!!

[Johnny Ma @_JohnnyMa_]

we really out here calculating P(pogchamp | gamer moment)

[modl.ai @modl_ai]

How engaging is your streamed game? It turns out that the behavior of your viewers can help us predict gameplay engagement with top accuracy! How? Check out our new @FDGconf paper “Moment-to-moment Engagement Prediction through the Eyes of the Observer” https://t.co/UUMiU6WWrn

[Johnny Ma @_JohnnyMa_]

sign me up

[Mario Klingemann💧💦 @quasimondo]

GTP-3 captured the gist of the typical art manifesto pretty well here. Though I still have no clear idea what "Void Art" might be. #gpt3

[sayna .*ೃ @OCEANGlRL]

crazy how kiera knightley has been acting since the late 1800's........ the grind never stops

[Johnny Ma @_JohnnyMa_]

"GPT-3 is capable of being highly original, it just doesn’t care about being original." - https://t.co/nwlF9jB7At

[Johnny Ma @_JohnnyMa_]

Markov chain generated texts are more fun to read than GPT-3 for this reason. No notion of good or bad, it just is.

[François Chollet @fchollet]

The way forward in AI is to focus on understanding, as opposed to mimicry.

If you want to understand the nature of life, you'd look at autopoiesis, molecular biology, DNA. If you wanted to mimic life, you'd be drawing cartoons. Big difference.

Lots of "AI" is a neural cartoon.

[Johnny Ma @_JohnnyMa_]

@Ted_Underwood We really need to run analysis on richer tags than "plot" and "mood"