Someone just made LLM weights absorb new knowledge without retraining. Researchers built In-Place Test-Time Training. It lets a model update its own weights while running. Not just during training. The trick: repurpose a matrix already inside every MLP block as "fast
RESEARCH
-
AI Kosh Launches Live Competitions for AI Innovation
By
–
Compete, build and win with AI !
— IndiaAI (@OfficialINDIAai) 17 avril 2026
AI Kosh now brings LIVE Competitions straight to the platform.
You have an excellent opportunity to showcase what you can build.
Explore to participate now: https://t.co/zbnejfjqlY#AIKosh #IndiaAI #AICompetitions #BuildWithAI #Innovation… pic.twitter.com/f45OowOfOfCompete, build and win with AI ! AI Kosh now brings LIVE Competitions straight to the platform. You have an excellent opportunity to showcase what you can build. Explore to participate now: https://
aikosh.indiaai.gov.in #AIKosh #IndiaAI #AICompetitions #BuildWithAI #Innovation -
Post-Quantum Readiness Starts Long Before Q-Day
By
–
Post-Quantum Readiness Starts Long Before Q-Day
by Etay Maor @Forbes Learn more: https://
bit.ly/4tdeZIu #EmergingTech #Innovation #Technology -
Eon Systems Copies Real Brain for Machine Intelligence
By
–
Most AI companies are trying to simulate intelligence.
— Pascal Bornet (@pascal_bornet) 17 avril 2026
Meet Dr. Alex Wissner-Gross, co-founder of Eon Systems, a startup trying to copy a real brain and run it inside a machine.
That is why this story is so hard to ignore.
Instead of building another artificial model, Eon took… pic.twitter.com/lJrJZ7fLZlMost AI companies are trying to simulate intelligence. Meet Dr. Alex Wissner-Gross, co-founder of Eon Systems, a startup trying to copy a real brain and run it inside a machine. That is why this story is so hard to ignore. Instead of building another artificial model, Eon took
-
Claude benchmark progress and capability evaluation across domains
By
–
The progress on some of these benchmarks has been insane! @AnthropicAI @DarioAmodei May I please ask you to request Claude to give you a list of the of the top 1000 areas of STEM, top 1000 magazine topics, top 500 professions, and for each list item pick a (not in training
-
Seedance 2.0 Advances Video Generation for World Complexity
By
–
Seedance 2.0: Advancing Generation for World Complexity Paper:
-
Microsoft Releases Memento Paper and Dataset for AI Research
By
–
find the paper and dataset here: https://
github.com/microsoft/meme
nto?tab=readme-ov-file
… -
Microsoft MEMENTO: LLM Reasoning Context Compression Method
By
–
Microsoft just mass-compressed LLM reasoning. their new paper introduces MEMENTO, a method that teaches reasoning models to manage their own context. instead of letting chain-of-thought grow into a flat 32K-token stream, the model learns to segment its reasoning into blocks,
-
Python Libraries for Data Science and Big Data
By
–
Python Libraries for #DataScience
by @Python_Dv #BigData #DataScientist -
Opus AI Model Performance Disappoints During Morning Testing
By
–
this sums it up for me: https://
x.com/kimmonismus/st
atus/2045059867925209329?s=20
… It's 10am in germany. I just started working 1.5hours ago and I was so negatively surprised by opus outputs. Wasnt really able to test Opus intensively yesterday evening.