.png "back to home page")
Nine months after saying it was withholding software called GPT-2 that generates text due to safety concerns, OpenAI just...tweeted it out. https://t.co/StNXEmvOjB
— Tom Simonite (@tsimonite) November 5, 2019
A thread
We're releasing the 1.5billion parameter GPT-2 model as part of our staged release publication strategy.
— OpenAI (@OpenAI) November 5, 2019
- GPT-2 output detection model: https://t.co/PX3tbOOOTy
- Research from partners on potential malicious uses: https://t.co/om28yMULL5
- More details: https://t.co/d2JzaENiks pic.twitter.com/O3k28rrE5l
Although in the end this report found no ill use of GPT2, and so the initial PR wave is still as silly in retrospect as it was then, from a skim I found the report to be pretty cool. Certainly nice to be proactive, too. And they included a model card on repo! New norms, nice. https://t.co/IRnECgdDQ4
— Andrey Kurenkov ? (@andrey_kurenkov) November 5, 2019
I wonder how the officially released 1.5B param GPT-2 model compares to the earlier 1.5B param OpenGPT-2 model preciously released by an independent effort to reproduce the results. ☝? https://t.co/xfEZeIDRAF
— hardmaru ? (@hardmaru) November 6, 2019
OpenAI has published the text-generating AI it said was too dangerous to share https://t.co/Jrg9Yuwh2K pic.twitter.com/7Qr3hXU01D
— The Verge (@verge) November 7, 2019
OpenAI has published the text-generating AI it said was too dangerous to share https://t.co/p0zpqwxrn6
— Martin Hawksey #altc (@mhawksey) November 7, 2019
OpenAI has published the text-generating AI it said was too dangerous to share https://t.co/ABL8Jh6GPB #AI pic.twitter.com/B2fVzvqSIM
— Reg Saddler (@zaibatsu) November 7, 2019
This news article about the full public release of OpenAI's 'dangerous' GPT-2 model was part written by GPT-2
— Iain Brown, PhD (@IainLJBrown) November 6, 2019
Read more here: https://t.co/mUE2QUeGxL#DataScience #MachineLearning #ArtificialIntelligence #NLP #AI #IoT #BigData
「悪用の可能性があり大変危険」としてOpenAIが公開を渋ってきた（そしてボコボコに批判された）テキスト生成モデルGPT-2の完全版モデル15億パラメータ版がついに公開https://t.co/eRaevAZ8dEhttps://t.co/sCDeheU3or
— えるエル (@learn_learning3) November 6, 2019
OpenAI's 'dangerous' #AI text generator is out: People find GPT-2's words 'convincing' https://t.co/2R7k9c1PyR via @ZDNet & @LiamT#AI #DigitalTransformation #Automation@labordeolivier
— Blockchain (@Blockchain_exp) November 7, 2019
@kalydeoo
@mallys_
@Ym78200
@3itcom
@CathCervoni
@lesguer_lionel
OpenAI's 'dangerous' #AI text generator is out:
— Spiros Margaris (@SpirosMargaris) November 7, 2019
People find #GPT2's words 'convincing' https://t.co/xlFlQ7nJBp #fintech #ArtificialIntelligence #CyberSecurity #MachineLearning #DeepLearning @ZDNet @LiamT @missdkingsbury @Shirastweet @helene_wpli @leimer @SabineVdL @YuHelenYu pic.twitter.com/2wH3xVgLDM
What would @NiemanLab look like if it was written by robots?
— Joshua Benton (@jbenton) November 7, 2019
I ran some of our stories through the new GPT-2 text-generation algorithm to see. Here are two:https://t.co/fXYFcfPYef pic.twitter.com/jOSQEdhoe2
Having apparently decided that the infopocalypse is not right around the corner, OpenAI has released its previously super-secret text-generation algorithm into the wild. The verdict, @jbenton? It's not about to take your job any time soon. https://t.co/wIB676kGWd @NiemanLab
— J Innovation City (@JInnovationCity) November 7, 2019