#Research
Curated coverage on «Research»: releases, research, benchmarks, open-source.
Hugging Face trained an image generation model in 24 hours
The third part of Hugging Face's PRX project shows that a full-fledged text-to-image model can be trained in just 24 hours. This changes perceptions of the…
Decima-8: the architecture seeking to reinvent neuromorphic chips
The new neuromorphic architecture Decima-8 proposes a radical approach: encoding signals with energy levels, digital crossbars instead of unreliable memristors, and…
AI helped formally verify a Fields Medal laureate’s proofs for the first time
The proofs of Ukrainian mathematician Maryna Viazovska on the sphere packing problem have undergone formal computer verification for the first time with AI…
Language model distillation: can knowledge theft be proven through chat
When machines decide what matters: AI searches for new physics in the collider data stream
A neural network as a time machine: why LLMs are taught to think the old-fashioned way
Goodbye, patches: TAPe + ML architecture changes the rules of computer vision
Unified Latents: Google DeepMind finds a way to improve AI generation
Google DeepMind's new Unified Latents framework solves the dilemma between data compression quality and neural network accuracy.
Sakana AI has learned to instantly adapt language models without fine-tuning
Tokyo-based lab Sakana AI introduced two methods — Doc-to-LoRA and Text-to-LoRA — that make it possible to adapt large language models to new data in fractions of a…
AI platform LUMI-lab sets record for gene-editing efficiency
The limits of scaling: why more AI agents do not guarantee results
DeepSeek, together with Tsinghua University and Peking University, improves AI agents' reasoning
Perplexity releases pplx-embed: embedding models that change the rules of search
Amazon Bedrock AgentCore: smart assistant for events
AWS showed how to deploy a production-ready AI agent with memory and personalization in a matter of hours — without writing custom infrastructure.
Andrej Karpathy fit GPT into 243 lines of pure Python
One of the most influential AI researchers showed that the entire transformer architecture fits into 243 lines of code without a single external library. The…