Exciting new AI research: "Transcendence: Generative Models Can Outperform The Experts That Train Them" shows LLMs trained on chess games from 1000 Elo players can achieve 1500 Elo performance! #AIResearch
MACHINE LEARNING
-
BAIR Researchers Win Major Awards at CVPR 2024
By
–
Congratulations to BAIR faculty and alumni for their awards at #CVPR2024 this week in Seattle. BAIR faculty @akanazawa won the Young Researcher Award. The R-CNN paper from @inkynumbers @jeffdonahue @trevordarrell @JitendraMalikCV received the Longuet-Higgins Prize for test of
-
Google Wins Best Paper Awards at CVPR 2026 Conference
By
–
@CVPR
, the premier computer vision conference happening now, has an amazing and diverse program with 2719 accepted papers (from 11352 submissions). It selected just two of these for the Best Paper Award, both with authors from @Google
: "Generative Image Dynamics"
All authors -
AI behaviors learned from examples, not explicit code
By
–
Those aren't coded. They're shown by examples, and some mysterious black box that mostly matches the examples becomes the AI.
-
Mastering Differential Privacy: Pure vs Approximate Methods
By
–
I had a dream last night where someone was asking whether pure or approximate differential privacy was better for situations with gross outliers and I (correctly) told them both would work. Now I can honestly say: I can answer that question in my sleep.
-
Domino Data Lab Named Gartner Magic Quadrant Visionary
By
–
Gartner shortlisted 18 data science and machine learning (DSML) vendors worldwide. Of those, @DominoDataLab emerged as one of the three named a Visionary in a highly competitive marketplace. Read our blog on the latest MQ: https://
domino.buzz/3VRhgKY #GartnerMQ #MLOps #AI -
Anthropic Sonnet 3.5 Shows 68% Improvement in Agentic Coding
By
–
Las pruebas internas realizadas por Anthropic muestran una mejora notable en tareas de "Agentic Coding". Es decir, tareas de programación que requieren ver un repositorio de código, hacer mejoras, arreglar fallos… Donde Opus resolvía el 38% Sonnet 3.5 ahora resuelve el 64%
-
Databricks Named Gartner Leader in Data Science ML Platforms
By
–
Databricks named a Leader in the 2024 Gartner® Magic Quadrant™ for Data Science and ML Platforms & positioned highest in Ability to Execute out of all evaluated vendors! We believe this recognition validates our focus on GenAI & traditional data science https://
dbricks.co/3VLPvU9 -
Deep Learning hits plateau: Has AI progress stalled?
By
–
La verdad… mi feed no luce como si el Deep Learning hubiera chocado contra un muro hace un par de años.
-
Four Domains of Systems: Communication, Computation, Classification, Control
By
–
You can read the work in many ways. The vertical axis represents time, and the horizontal addresses systems in four domains: communication, computation, classification and control. You can follow a theme, a time period, or a set of ideas in any direction.