In this tutorial, we write a mini reinforcement learning setup in which a multi-agent system learns to navigate a grid world through interaction, feedback, and hierarchical decision-making. We built everything from the ground up...
Nano Banana Pro, also known as Gemini 3 Pro Image, is a new image generation and editing model built by Google DeepMind on Gemini 3 Pro. It is positioned as a state-of-the-art image creation...
How can teams run tera-parameter language models on existing hybrid GPU clusters without expensive new hardware or deep vendor lock-in? Perplexity’s research team released TransferEngine and the surrounding pplx Garden toolkit as an open...
In this tutorial, we implemented a complete workflow for building, tracking, and evaluating LLM pipelines using opik. We build the system step by step, starting with lightweight models, adding prompt-based planning, creating datasets, and...
The Allen Institute for Artificial Intelligence (AI2) is releasing Olmo 3 as a fully open model series that exposes the entire “model flow,” from raw data and code to intermediate checkpoints and deployment-ready variants....
In this tutorial, we explore how to build a fully offline, multi-step inference agent that uses lecturer Library to generate structured output and reliably orchestrate complex tool calls. In this implementation, we design an...
How to reliably find, segment, and track every instance of any concept in a large collection of images and videos using simple tips? The Meta AI team has just released Meta Segment Anything Model...
Focus more on the “why” of advertising… In my last few videos, I talked about how advertisers spend too much time worrying about the “how.” When they don’t get results, they obsess over...
Producing LLM services is now a system issue rather than a generate() ring shape. For real workloads, the choice of inference stack affects your Tokens per second, tail latency,final Cost per million tokens on...
OpenAI introduces GPT-5.1-Codex-Max, a cutting-edge agent coding model designed for long-running software engineering tasks involving millions of tokens and multi-hour sessions. It is now available in the CLI, IDE extension, cloud integration, and code...
Google launched Antigravity as an agent development platform on top of Gemini 3. It’s not just an autocomplete layer, it’s an IDE where agents can plan, execute, and interpret complex software tasks across editor,...
In this tutorial, we take a deep dive into how to systematically benchmark agent components by evaluating multiple inference strategies on different tasks. We explore how different architectures (such as Direct, Chain-of-Thought, ReAct, and...
In this tutorial, we build an advanced agent deep reinforcement learning system that guides the agent to not only learn actions in the environment, but also learn how to choose its own training strategy....