Intellectual hub of the topic
Midjourney has released the alpha version of V8. This update improves image quality, control, and generation speed.
Cursor took an unusual approach to training its AI assistant: self-summarization became part of the training process, not just a supporting tool.
AI: Events
Technical context • Infrastructure
The Mixture of Experts approach allows language models to work more efficiently by activating only a portion of their capabilities for each specific task.
AMD explains how to easily deploy the Qwen3-5 language model on its Developer Cloud service using the SGLang framework.
AI: Events
Research
Researchers have introduced M4-RAG, a large-scale benchmark for evaluating systems that answer questions about images by drawing on external knowledge and operating in multiple languages.
AI consumes vast amounts of electricity and fuels global warming, yet it also offers solutions. We'll explore whether this is an absurdity or a genuine chance to escape the vicious cycle.
AI: Events
Infrastructure
Exploring how Kimi created a scalable and secure environment for AI agents on the Alibaba Cloud platform.
What if you could train a massive neural network using half the memory – without breaking anything? That's exactly what the creators of FlashOptim are exploring.
NXP and Hugging Face explain how to train robotic artificial intelligence on custom data and run it on a low-power embedded device.
Don’t miss a single experiment!
Subscribe to our Telegram channel —
we regularly post announcements of new books, articles, and interviews.