My latest story for @WIRED was a fun one to report. A bot called Nexto is tearing up Rocket League, and it uses AI that was only found in top research labs a few years ago. It shows how advances in AI are spreading, and how online games will need to adapt.
RESEARCH
-
Comprehensive Performance Comparison of YOLO Object Detection Models
By
–
With the number of YOLO models, it's tough to choose when working on object detection projects. But by running a thorough performance comparison, it becomes a lot easier. That’s precisely what we do in our blog post. https://t.co/92yGeRRp2m#yolov5 #yolov6 #yolov7 #gpu #ai pic.twitter.com/K7ziMpRfeJ
— Satya Mallick (@LearnOpenCV) 19 janvier 2023With the number of YOLO models, it's tough to choose when working on object detection projects. But by running a thorough performance comparison, it becomes a lot easier. That’s precisely what we do in our blog post. https://
learnopencv.com/performance-co
mparison-of-yolo-models/
… #yolov5 #yolov6 #yolov7 #gpu #ai -
@peterwildeford notes GPT-3 poor at BIG-bench ASCII word recognition
By
–
As @peterwildeford points out a version of this task is in BIG-bench and GPT-3 is really bad at it: https://
github.com/google/BIG-ben
ch/tree/main/bigbench/benchmark_tasks/ascii_word_recognition
… -
AI says ‘mountain’ frequently, even for ASCII smiley face art
By
–
It says “mountain” a lot though. I gave it ASCII art of a smiley face and it said “mountain” too. I suspect it has some intuition about shapes being vaguely square vs. triangular etc. but mostly I think it’s good at guessing what typical subjects are.
-
OpenAI Prepares Text-to-Video Model Launch
By
–
After ChatGPT (text-to-text) and DALL.E (text-to-image), get ready for the new text-to-video model by OpenAI It is a legitimate research project at OpenAI and could launch much sooner than anticipated!!
-
Reinforcement Learning Model Challenged: New Understanding Emerges
By
–
Reinforcement learning was wrong all along. Now we know why, and how to do better. https://
economist.com/science-and-te
chnology/2023/01/18/a-decades-old-model-of-animal-and-human-learning-is-under-fire
… -
OpenAI Dactyl uses ADR to solve a Rubik’s Cube
By
–
2. OpenAI Dactyl project
— AI Breakfast (@AiBreakfast) 19 janvier 2023
Automatic Domain Randomization (ADR) for robotics endlessly generates progressively more difficult environments in simulation. ADR exposes the neural network to many different variants of the same problem, in this case solving a Rubik’s Cube.
(cont.) pic.twitter.com/3Vna5HWNVV2. OpenAI Dactyl project Automatic Domain Randomization (ADR) for robotics endlessly generates progressively more difficult environments in simulation. ADR exposes the neural network to many different variants of the same problem, in this case solving a Rubik’s Cube. (cont.)
-
OpenAI training with ADR for sim-to-real Rubik’s cube solving
By
–
Using ADR, OpenAI trains with exposure to complexity which prepares the network to transfer from simulation to the real world since it has to learn to quickly identify and adjust to whatever physical world it is confronted with.
— AI Breakfast (@AiBreakfast) 19 janvier 2023
More on the project here: https://t.co/7m2EbrUEkK pic.twitter.com/dFbn3lmBDLUsing ADR, OpenAI trains with exposure to complexity which prepares the network to transfer from simulation to the real world since it has to learn to quickly identify and adjust to whatever physical world it is confronted with. More on the project here: https://
openai.com/blog/solving-r
ubiks-cube/
… -
Machine Learning Protects Sharks While Minimizing Fisheries Impact
By
–
Even apex predators face risks. Great read in @FrontMarineSci on how researchers @emLabUCSB are using #machinelearning to map where #sharks can be protected with minimal impact to #fisheries. #randomforestmodel http://
ow.ly/j4lz50MuAPK -
What Happens When AI Has Read Everything Online?
By
–
What Happens When #AI Has Read Everything? https://
theatlantic.com/technology/arc
hive/2023/01/artificial-intelligence-ai-chatgpt-dall-e-2-learning/672754/
… @TheAtlantic #MachineLearning #BigData #Analytics @Shi4Tech @CatherineAdenle #100DaysofCode #womenwhocode #serverless #ML @anijov #IoT @FmFrancoise @BIScorecard #DataScience #DeepLearning #DigitalTransformation