LLaDA2.0-Uni Unifying Multimodal Understanding and Generation with Diffusion Large Language Model paper: https://
huggingface.co/papers/2604.20
796
…
MACHINE LEARNING
-
LLaDA2.0-Uni: Unified Multimodal Understanding and Generation with Diffusion
By
–
-
ICLR 2026 Outstanding Papers Selection Procedure Announced
By
–
More details on the recognized papers and the selection procedure here: https://
blog.iclr.cc/2026/04/23/ann
ouncing-the-iclr-2026-outstanding-papers/
… -
ICLR 2026 Outstanding Paper Award Winners Announced
By
–
Congratulations to the winners of the #ICLR2026 Outstanding Paper Award! It was an honour to chair the committee, the entire team did a great job. We recognized two outstanding papers and one honorable mention. Check out the blog post for more: https://
blog.iclr.cc/2026/04/23/ann
ouncing-the-iclr-2026-outstanding-papers/
… -
Autogenesis: AI agents rewrite themselves without human help
By
–
AI agents can now rewrite themselves without human help. Most AI agents stop improving the moment they ship. New tools arrive, environments shift, and the agent stays frozen in time. ' Retraining is expensive and human patches are brittle. A new paper called Autogenesis
-
Voice AI Accuracy Gap: Async Benchmarks 81.2% Sentence-Level Performance
By
–
The accuracy gap in production voice AI is wider than most people realize. Async’s benchmark on non-standard text shows 81.2% sentence-level accuracy and 88.6% unit-level accuracy, tested on 2,200+ non-standard words in true real-time streaming. No preprocessing. No text
-
Tesla FSD Advances Decision-Making Over Vision Capabilities
By
–
The most important shift in driving AI may not be that cars see more.
— Pascal Bornet (@pascal_bornet) 23 avril 2026
It is that they are starting to understand more.
What stands out to me is the quality of the decision.
In this clip, Tesla’s Full Self-Driving appears to recognize an approaching emergency vehicle and begin… pic.twitter.com/pQDw5qPBSZThe most important shift in driving AI may not be that cars see more. It is that they are starting to understand more. What stands out to me is the quality of the decision. In this clip, Tesla’s Full Self-Driving appears to recognize an approaching emergency vehicle and begin
-
AI Labs Shift Focus to Smarter Pretraining Over Test-Time Reasoning
By
–
Quick update to GPT-5.5 / Spud: AI labs are shifting back toward making models smarter during pretraining rather than leaning on reasoning (test-time compute) to boost performance. OpenAI's Spud and Anthropic's Mythos both appear to reflect this trend, getting better answers
-
Scaling Agent RL with Real-World Environment Synthesis
By
–
"Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence" This paper scales agent RL with actual stateful tool environments instead of toy simulators. A key component is that the environment set is not frozen but continuously expanded
-
Cutting-edge AI capabilities revolutionize specialized task performance
By
–
It's amazing what AI can tell us. This is bleeding edge AI, better at doing this kind of stuff than any other.
-
Sony’s Ace Robot Achieves Expert-Level Ping-Pong Play
By
–
Incredible work by Sony published in @Nature today! 🏓
— hardmaru (@hardmaru) 23 avril 2026
They’ve built “Ace”, an autonomous ping-pong robot that uses RL and Sony’s vision sensors to achieve expert-level play in ping pong. A huge leap forward for adaptive robotics.https://t.co/hJqnZzXV17pic.twitter.com/xQ5NThU6xJ https://t.co/hpep1aQsDBIncredible work by Sony published in @Nature today! They’ve built “Ace”, an autonomous ping-pong robot that uses RL and Sony’s vision sensors to achieve expert-level play in ping pong. A huge leap forward for adaptive robotics. https://
nature.com/articles/s4158
6-026-10338-5
…