NeurIPS attendance (real + virtual) is down by more than a third this year.
RESEARCH
-
GPT-3 succeeds on trick questions, pre-training unchanged since 2021
By
–
No. It also succeeds on questions I’ve written myself that trick GPT-3, as in this screenshot. The pre-training data hasn’t been updated since 2021, and OpenAI specifically denies (in this thread) training on the Hofstadter/Bender questions.
-
Meta AI Thanks Visitors at NeurIPS 2022 Booth
By
–
This week has been flying by — thanks to everyone who has stopped by the booth to chat or check out our demos at #NeurIPS2022!
-
DeepMind Reduces Math Reasoning Errors Using Process-Based Supervision
By
–
DeepMind Studies Process- vs Outcome-based Model Supervision, Significantly Reducing Reasoning Errors on Math Word Problems https://
syncedreview.com/2022/11/30/dee
pmind-studies-process-vs-outcome-based-model-supervision-significantly-reducing-reasoning-errors-on-math-word-problems/
… -
ChatGPT reduces hallucinations, avoids false-assumption pitfalls
By
–
To be fair, they don’t claim to have solved hallucination fully, and explain why in the announcement. But it’s clearly better, and doesn’t seem to fall for these false-assumption questions that trick GPT-3. https://
openai.com/blog/chatgpt/ -
Google Phenaki: Text-to-Video Generation Breakthrough Presentation
By
–
Today at 3:30 pm @RubenEVillegas and @piekindermans will be at the Google booth to describe how the Phenaki model overcomes the challenges of text-to-video generation to synthesize realistic videos from text prompts. More at: https://t.co/B2OdMk8l2S pic.twitter.com/f0KohBFnbF
— Google AI (@GoogleAI) 30 novembre 2022Today at 3:30 pm @RubenEVillegas and @piekindermans will be at the Google booth to describe how the Phenaki model overcomes the challenges of text-to-video generation to synthesize realistic videos from text prompts. More at: https://
phenaki.research.google -
Health Monitoring for Large-Scale AI URL Datasets
By
–
We should have a health page with updated indicators for urls datasets like LAION or COYO 83% █ █ █ █ █ █ □ □ Self destructing datasets!
-
Meta AI’s CICERO Masters Diplomacy Game at Human Level
By
–
Meta AI’s CICERO is the first AI to play Diplomacy at a human level. This intricate strategy game is about people rather than pieces since to win, players must build trust, cooperate and negotiate with others. Learn more about #CICERObyMetaAI's gameplay
-
ChatGPT defeats Hofstadter/Bender hallucination questions
By
–
OpenAI’s new ChatGPT appears to defeat Hofstadter/Bender’s list of hallucination-inducing questions, published in The Economist this June to demonstrate the “hollowness” of GPT-3’s understanding of the world: https://
economist.com/by-invitation/
2022/06/09/artificial-neural-networks-today-are-not-conscious-according-to-douglas-hofstadter
… -
Cloud Infrastructure Challenges for Large-Scale AI Model Development
By
–
“Many companies would like to be able to quickly research and develop these large-scale models, but the infrastructure that exists in traditional cloud just doesn't make this kind of large-scale R&D easy," says Andy Hock. Read more in Tiernan Ray's article