علوم
2872 مقال
Language models can explain neurons in language models
We use GPT-4 to automatically write explanations for the behavior of neurons in large language models and to score those explanations. We release a dataset of these (imperfect) explanations and scores for every neuron in GPT-2.
Upholding Our Commitment to Protecting Your Privacy: What the FTC Gets Wrong - meta.com
Upholding Our Commitment to Protecting Your Privacy: What the FTC Gets Wrong meta.com
Scientists film deepest ever fish on seabed off Japan
Cruising at a depth of 8,336 meters (over 27,000 feet) just above the seabed, a young snailfish has become the deepest fish ever filmed by scientists during a probe into the abyss of the northern Pacific Ocean.
The four astronauts NASA picked for the first crewed moon mission in 50 years
Astronauts who will helm the first crewed moon mission in five decades were revealed on Monday, queuing up the quartet to begin training for the historic Artemis II lunar flyby that is set to take off in November 2024.
One Medical joins Amazon to make it easier for people to get and stay healthier - About Amazon
One Medical joins Amazon to make it easier for people to get and stay healthier About Amazon
How AWS will return more water than it uses by 2030 - About Amazon
How AWS will return more water than it uses by 2030 About Amazon
Our Largest Ever Climate Survey Can Inform Policies, Research and Campaigns Around the World - meta.com
Our Largest Ever Climate Survey Can Inform Policies, Research and Campaigns Around the World meta.com
Scientists discover nine new coronaviruses in 11 days using AWS - About Amazon
Scientists discover nine new coronaviruses in 11 days using AWS About Amazon
Artificial intelligence used to recognise primate behaviours in the wild - University of Oxford
Artificial intelligence used to recognise primate behaviours in the wild University of Oxford
Amazon's Sustainability Data Initiative helps advance climate research - About Amazon
Amazon's Sustainability Data Initiative helps advance climate research About Amazon
Tackling climate change, one forest at a time - About Amazon
Tackling climate change, one forest at a time About Amazon
Multimodal neurons in artificial neural networks
We’ve discovered neurons in CLIP that respond to the same concept whether presented literally, symbolically, or conceptually. This may explain CLIP’s accuracy in classifying surprising visual renditions of concepts, and is also an important step toward understanding the associations and biases that CLIP and similar models learn.
Multimodal neurons in artificial neural networks
We’ve discovered neurons in CLIP that respond to the same concept whether presented literally, symbolically, or conceptually. This may explain CLIP’s accuracy in classifying surprising visual renditions of concepts, and is also an important step toward understanding the associations and biases that CLIP and similar models learn.
Stepping Up the Fight Against Climate Change - meta.com
Stepping Up the Fight Against Climate Change meta.com
Astrophotography: Stunning images of Idaho's night sky
Sunday Morning Astrophotography: Stunning images of Idaho's night sky .chip { background-image: url('/fly/bundles/cbsnewscore/images/chip-bgd/chip-bgd-sunday-morning.jpg'); } By David Morgan September 13, 2020 / 6:00 AM EDT / CBS News Add CBS News on Google Jordan Ragsdale A view of the Veil Nebula in the constellation Cygnus.The Central Idaho Dark Sky Reserve, draped over nearly one million acres of wilderness areas and the Sawtooth National Forest, provides stargazers a rare opportunity to enj...
A Second Update on Our Civil Rights Audit - meta.com
A Second Update on Our Civil Rights Audit meta.com
Reinforcement learning with prediction-based rewards
We’ve developed Random Network Distillation (RND), a prediction-based method for encouraging reinforcement learning agents to explore their environments through curiosity, which for the first time exceeds average human performance on Montezuma’s Revenge.
Learning complex goals with iterated amplification
We’re proposing an AI safety technique called iterated amplification that lets us specify complicated behaviors and goals that are beyond human scale, by demonstrating how to decompose a task into simpler sub-tasks, rather than by providing labeled data or a reward function. Although this idea is in its very early stages and we have only completed experiments on simple toy algorithmic domains, we’ve decided to present it in its preliminary state because we think it could prove to be a scalable a...
Learning complex goals with iterated amplification
We’re proposing an AI safety technique called iterated amplification that lets us specify complicated behaviors and goals that are beyond human scale, by demonstrating how to decompose a task into simpler sub-tasks, rather than by providing labeled data or a reward function. Although this idea is in its very early stages and we have only completed experiments on simple toy algorithmic domains, we’ve decided to present it in its preliminary state because we think it could prove to be a scalable a...
OpenAI Scholars 2019: Applications open
We are now accepting applications for our second cohort of OpenAI Scholars, a program where we provide 6–10 stipends and mentorship to individuals from underrepresented groups to study deep learning full-time for 3 months and open-source a project.