All streams

I constantly have multiple Chrome windows open, each with a pile of tabs. One window has a dozen slow analytics queries that I'll check "any minute now." Another has my research on $lookup and $unwind in MongoDB. The third one — with the most tabs — has local school enrollment rules, because life.

Tab-saving extensions have existed for ages, but reviews regularly complain about data loss — so why not build my own, designed specifically for backup? Every Chrome user already has a Google account, usually with sync enabled.

Here's how I used it to make Tab Saver extension that backs up to your Google Account without any signups and with zero configuration needed. It's pretty straightforward - you can easily use it as well.

Much attention is currently focused on the size of neural networks and the gigawatts of power consumed by data centers. However, the future lies not only in giant clusters but also in tiny chips embedded directly into the sensing elements of hardware. When a neural network is placed directly inside a sensor chip, it must be exceptionally efficient.

Through experimentation, I have successfully built a neural network architecture with 380 parameters (with potential for further reduction), capable of operating in conditions considered unsuitable for conventional algorithms.

This guide covers how to set up OpenClaw (formerly Clawdbot) on your local machine and, most importantly, how to secure it so strangers can’t access your computer. If you are ready, then let’s get started! :)

It’s surprising that there’s almost no practical information about Telegram Cocoon beyond what’s on GitHub and the official website. Various media outlets have plenty of general coverage about the network launch, but almost nothing about real user experience.

I decided to spend a bit of time and figure out what’s actually going on in the network, how it works, and, most importantly, whether I, as a developer, can use it today. So in this article I’ll look at Cocoon from a developer’s perspective: how to install it and how to use it.

Привет, Хабр! Меня зовут Юра Петров, я руководитель отдела разработки компании Friflex и автор канала «Мобильный разработчик». OpenAI на днях выпустила крутую статью, где впервые подробно описала работу своего агента для написания и изменения кода — Codex CLI. Сердце системы — «агентский цикл» (agent loop).

Это процесс, в котором модель получает задачу от пользователя, при необходимости вызывает инструменты (например, запускает команды в терминале), анализирует результат и повторяет цикл, пока не получит финальный ответ или не внесет нужные изменения в код. Статья фокусируется на том, как устроен этот цикл, как формируются запросы к модели и как система управляет контекстом.

Asynchronous I/O, ML-based query plan optimization, and built-in connection pooling are among the key features of the new Postgres Pro Enterprise 18. This release brings together the capabilities of the vanilla PostgreSQL 18 core with Enterprise-grade tools for working with large-scale data. Today we will walk through the technical details, new index scanning strategies, and mechanisms for scaling write workloads.

Trends in UX design change rapidly – not even by year, but rather by month. As smartphones and technologies constantly evolve, so do UX rules – that's why a product that was perfectly crafted according to the last year's trends most likely looks outdated already and needs a thorough update. Some trends stand the test of time, though, while others get quickly replaced by more relevant ones. Here are top 10 trends that are going to shape UX design in 2026:

Do you switch between Linux and Windows in dual-boot? Then you're probably familiar with this problem: you have to reconnect all your Bluetooth devices every time. Headphones, mouse, keyboard, gamepad — everything has to be reconnected.

It's scary to even think about it:
3 devices × 90 seconds × 3 switches per day × 250 days = 56 hours wasted per year.

I spent a month solving this problem and wrote BlueVein — a utility for automatically synchronizing Bluetooth keys between operating systems.

In modern neural networks, including Transformer-based LLMs, unbounded activation functions—ReLU and GELU—have become the standard. Their main advantage is good gradient flow and the rapid training of deep models.

However, in practice, a problem is observed: when dominant patterns or high-frequency noise appear in the input context (long dialogues, noisy data, repetitive or dominant tokens), models become unstable and prone to generation degradation and hallucinations.

In this article, I attempted to find out if the choice of activation function could be fundamentally linked to LLM hallucinations.

In my previous article, I noted some interesting behavior regarding Weight Decay; here, I examine it in detail.

It is generally accepted in the ML industry that if we take a pre-trained model and fine-tune it on a new task, the old weights are gradually overwritten. Furthermore, if we add Weight Decay (L2 regularization), the process of "forgetting" superfluous information should theoretically happen even faster.

I tested this claim experimentally. The results were counter-intuitive: under specific settings, Weight Decay works in the exact opposite way—it protects the old structure from destruction.

Below is a description of the experiment and conclusions for those involved in model training and AI safety.

In January 2026, Ollama added support for the Anthropic Messages API, enabling Claude Code to connect directly to any Ollama model. This tutorial explains how to install Claude Code, pull and run local models using Ollama, and configure your environment for a seamless local coding experience.

In my previous article, I explored the phenomenon of subliminal learning, but it raised more questions than answers. It is time to dive deeper. Below, you will find the experiments and the code.

In the fields of AI Alignment and LLM Security, a critical question remains: does fine-tuning or Reinforcement Learning from Human Feedback (RLHF) guarantee the removal of unwanted information?

Spoiler: The experiments demonstrated that the well-known Mode Connectivity effect makes the complete erasure of pre-training information practically impossible during standard fine-tuning. Structural Imprinting persists in the weight topology and can be read through a subliminal channel. Even with full weight unfreezing and aggressive L2 regularization (active forgetting), the latent space topology formed during the pre-training stage persists and determines the solution to the new task with an accuracy of 88–99%.

Today we want to talk about choosing a DBMS for WMS not as a dry technical discussion, but as a strategic decision that determines the security, budget, and future flexibility of your business. This is not about "why PostgreSQL is technically better," but about why it has become the only safe, cost-effective, and future-proof solution for Russian warehouse systems in the new reality.

This is not just another database article. It is a roadmap for those who do not want to wake up one day with a paralyzed warehouse and multi-million fines due to a bad decision made yesterday. At INTEKEY we have gone this path deliberately, and today our WMS projects for the largest market players run on PostgreSQL. We know from experience where the pitfalls are and how to avoid them.

I was so tired of being the worst at emojis on Google Meet that I reverse engineered its WebRTC messages to create a Google Chrome extension that shows the most popular emojis in the team and allows you to search by meaning and with typos.

Recently, I started using Session Teleportation in Claude Code. It allows you to move an entire conversation, including context, history, and the working branch, between the web and your local terminal.

In this tutorial, I show you how it works and how to use it to make your workflow seamless.

Modern neural network training often resembles alchemy. We have working recipes, but how exactly a statistical model transforms terabytes of text into understanding remains unclear.

Why is subliminal learning (pattern transmission through noise) possible? Why does training on synthetic data lead to degradation, even when the data appears to be of high quality?

In this article, I propose looking at training architecture from a different angle. The core idea is simple: positive definitions in high-dimensional space are computationally inefficient. A neural network does not learn what an object is. It learns what the object is not, and the model's intelligence depends entirely on the quality of this "NOT."

What follows is the theory, experiments in PyTorch (code included), mathematics, and an explanation of why LLM collapse is highly probable.

Bulk config rollouts, built-in OpenTelemetry, and two-click HA cluster control are all part of one goal: making PostgreSQL admin simpler and safer. PPEM 2.3 is a big step toward that — with user-defined presets, a reworked alerting system, and stronger RBAC — helping you bring order to messy configs and trust the system to warn you before things go sideways.

If you are using the Codex CLI and find yourself writing the same instructions over and over again, you are not using the tool to its full potential. Codex offers a powerful feature called Skills that allows you to package reusable workflows and give your AI agent new capabilities on demand. If you want to know more about it, then read this article until the end.

As we wrap up another year, it's time to look back at what our department has accomplished. 2025 brought us 42 published papers spanning fundamental ML theory, applied AI systems, and cutting-edge optimization methods—from transformer Hessians and generative models to hallucination detection and matrix-oriented optimizers.

Beyond publications, our students won competitions and defended their theses: 14 Bachelor's, 9 Master's, 3 PhD, and 1 DSc dissertations. They also launched ambitious group research projects. Three of our faculty and alumni received the prestigious Yandex ML Prize, and our head Konstantin Vorontsov was inducted into the Hall of Fame. If you read our summer overview of thesis defences or last winter's year-in-review for 2024, this post continues that story with the next chapter.

In this year-in-review, we dive into the research highlights, share stories from our educational programs, and celebrate the community that makes it all possible.

A common task in enterprise systems is to load large volumes of data into PostgreSQL — sometimes tens or even hundreds of millions of rows. At first glance, this seems simple: just write a loop in Java and call save() for every record. But in reality, such an approach can be painfully slow. Even a perfectly tuned PostgreSQL instance won’t help if the application is sending data inefficiently.

This article explains how to significantly accelerate bulk inserts when working with PostgreSQL through Spring and Hibernate. We’ll walk through which Spring and Hibernate settings are worth enabling, why they matter, and how much performance they can actually unlock. We’ll also look at how to build your own data-insertion layer for PostgreSQL — one that lets you switch between different insertion strategies, leverage PostgreSQL’s custom capabilities, and parallelize the process. Finally, we’ll see how to integrate this layer with Spring and what real gains each approach can deliver.