I believe there were no ill intentions with Reflection 70B; they would gain nothing by lying. Instead, it's a lesson in caution: scrutinize AI developments, uphold transparency, and let the community validate claims.
RESEARCH
-
BAIR Faculty Wins Max Planck Humboldt Medal for Computational Microscopy
By
–
Congratulations to BAIR faculty @optrickster who has been awarded the Max Planck Humboldt Medal "…for her work in computational microscopy…"
-
Neural Operators Beyond Transformers for Scientific Modeling
By
–
No. Not all are transformers. Extensions to Neural Operators are key to scientific modeling.
-
Code Data Improves LLM Generalization Beyond Coding Tasks
By
–
New from @CohereForAI and @cohere
! To Code, or Not To Code? This paper reveals that including code data in LLM pre-training is critical for generalization far beyond coding tasks. The authors @viraataryabumi @ahmetustun89 and @sarahookr are here to answer your questions! -
AI Accelerating Scientific Discovery in Biology and Weather
By
–
I learned so much from @AnimaAnandkumar when she came to visit my class at Stanford. Her insights about using AI to accelerate scientific discovery in biology, weather and other areas were fascinating.
-
AI Weather Forecasting with Neural Operators and LLMs
By
–
I was honored to be one of the speakers at @erikbryn
's class on AI and creativity at @StanfordHAI last Spring to discuss the future of AI, LLMs, and Neural Operators. How we got to build the first AI-based weather model FourCastNet using Neural Operators, the first AI agent -
NeuralGCM: 100,000x Faster AI Climate Simulation Model
By
–
Simulating a year of Earth's atmosphere with the state-of-the-art X-SHiELD model is slow and costly. NeuralGCM is an innovative hybrid #AI model that is 100,000 times more computationally efficient. Learn more about this groundbreaking technology: https://t.co/UqPj6x2HW7 pic.twitter.com/k0TtyefBxq
— Google AI (@GoogleAI) 10 septembre 2024Simulating a year of Earth's atmosphere with the state-of-the-art X-SHiELD model is slow and costly. NeuralGCM is an innovative hybrid #AI model that is 100,000 times more computationally efficient. Learn more about this groundbreaking technology: https://
goo.gle/4egSGK2 -
ScribblePrompt Outperforms SAM in Neuroimaging Annotation
By
–
ScribblePrompt reduced annotation time by 28% compared to Meta’s Segment Anything Model (SAM). Its self-correcting, interactive capabilities made ScribblePrompt the preferred tool among neuroimaging researchers at MGH in a user study. 93.8% of these users favored the MIT
-
ScribblePrompt: AI-Powered Interactive Image Segmentation Tool
By
–
Authors: Hallee Wong (
@HalleeWong
), Marianne Rakic (
@MarianneRakic
), John Guttag, and Adrian Dalca (
@AdrianDalca
)
Paper: https://
arxiv.org/abs/2312.07381 Website: https://
scribbleprompt.csail.mit.edu
Video: https://
bit.ly/3Tn76Ac -
ScribblePrompt: AI Algorithms Simulate Human Annotation in Medical Imaging
By
–
To train ScribblePrompt, the team used algorithms to simulate how humans would scribble & click on different regions in medical images. In addition to commonly labeled regions, the team also used superpixel algorithms to identify potential new regions of interest to medical