What if you could boost any AI's performance just by rephrasing your prompts? Researchers from FaceMind & CUHK propose "Adam's Law": a simple but powerful principle that more common, frequently seen text improves LLMs. Their method paraphrases inputs into more frequent
@jiqizhixin
-
Image Generators Emerge as Generalist Vision Learning Models
By
–
Image Generators are Generalist Vision Learners Paper: https://
arxiv.org/abs/2604.20329
Project: https://
vision-banana.github.io -
Google Vision Banana: Instruction-Tuned Generalist Image Generator Model
By
–
Huge! Google just proved Image Generators are Generalist Vision Learners! They introduce Vision Banana, a model built by instruction-tuning a base image generator (Nano Banana Pro). Instead of using special systems for different tasks, they reframe every vision problem—like
-
Sol-RL: Train AI Image Generators 4x Faster
By
–
What if you could train AI image generators 4x faster without losing quality?
— 机器之心 JIQIZHIXIN (@jiqizhixin) 23 avril 2026
NVIDIA, HKU, and MIT researchers present Sol-RL.
They use a clever two-stage process: first, a super-fast, low-precision search creates a huge pool of image candidates. Then, they regenerate only the… pic.twitter.com/KokLk4tmr9What if you could train AI image generators 4x faster without losing quality? NVIDIA, HKU, and MIT researchers present Sol-RL. They use a clever two-stage process: first, a super-fast, low-precision search creates a huge pool of image candidates. Then, they regenerate only the
-
DeepSeek Releases Tile Kernels GPU Optimization for LLMs
By
–
DeepSeek just released Tile Kernels! Tile Kernels are optimized GPU kernels for LLM operations, built with TileLang. DeepSeek claims that “most kernels in this project approach the limits of hardware performance in terms of compute intensity and memory bandwidth. Some of them
-
OmniRoam: AI Generates Endless Panoramic Videos from Photos
By
–
What if you could wander endlessly through a world created from just a single photo or short video?
— 机器之心 JIQIZHIXIN (@jiqizhixin) 23 avril 2026
A team from UC Irvine, UC San Diego, and Adobe Research presents OmniRoam.
It’s a new AI that generates long, panoramic videos that let you explore a scene. You give it a… pic.twitter.com/3jPsuR4PvaWhat if you could wander endlessly through a world created from just a single photo or short video? A team from UC Irvine, UC San Diego, and Adobe Research presents OmniRoam. It’s a new AI that generates long, panoramic videos that let you explore a scene. You give it a
-
DataFlex: Dynamic Data Optimization for LLM Training
By
–
What if you could supercharge LLM training by dynamically optimizing the data, not just the model? Researchers from Peking University, Shanghai AI Lab, & the LLaMA-Factory team present DataFlex. It's a unified framework that smartly selects, mixes, and re-weights training data
-
Process Reward Models Grade Robot Performance Like Sports Replay
By
–
What if we could audit a robot's performance like a sports replay, grading every move, not just the final score? Researchers from Peking University, Chinese Academy of Sciences, and the Beijing Academy of AI present PRM-as-a-Judge. They use a "Process Reward Model" to watch a
-
LatentUM: AI Model Processes Images Text Actions Simultaneously
By
–
What if an AI could think in pictures and words simultaneously, without the usual translation lag? Researchers from Shanghai Jiao Tong U, Tsinghua U, and UCSD present LatentUM. They built a single model that processes images, text, and actions all in one shared "semantic
-
Beyond Words: AI’s Shift to Latent Space Reasoning
By
–
What if the next leap in AI isn't about better words, but moving beyond words entirely? A massive team from NUS, Tsinghua, Tencent, & others presents a unified survey on the "Latent Space." They argue AI's core reasoning is shifting from explicit token-by-token generation to