How can we get large model level multimodal reasoning for documents, charts and videos while running only a 3B class model in production? Baidu has added a new model to the ERNIE-4.5 open source family. ERNIE-4.5-VL-28B-A3B-Thinking is a vision language model that focuses on document, chart and video understanding with a small active parameter budget….
def generate_advanced_dataset(): np.random.seed(42) start_date = datetime(2022, 1, 1) dates = [start_date + timedelta(days=x) for x in range(730)] categories = [‘Electronics’, ‘Clothing’, ‘Home & Garden’, ‘Sports’, ‘Books’] products = { ‘Electronics’: [‘Laptop’, ‘Smartphone’, ‘Headphones’, ‘Tablet’, ‘Smartwatch’], ‘Clothing’: [‘T-Shirt’, ‘Jeans’, ‘Dress’, ‘Jacket’, ‘Sneakers’], ‘Home & Garden’: [‘Furniture’, ‘Lamp’, ‘Rug’, ‘Plant’, ‘Cookware’], ‘Sports’: [‘Yoga Mat’, ‘Dumbbell’, ‘Running Shoes’,…
How do you build a single speech recognition system that can understand 1,000’s of languages including many that never had working ASR (automatic speech recognition) models before? Meta AI has released Omnilingual ASR, an open source speech recognition suite that scales to more than 1,600 languages and can be extended to unseen languages with only…
In this tutorial, we explore how to build and train an advanced neural network using JAX, Flax, and Optax in an efficient and modular way. We begin by designing a deep architecture that integrates residual connections and self-attention mechanisms for expressive feature learning. As we progress, we implement sophisticated optimization strategies with learning rate scheduling,…
Modern agentic applications rarely talk to a single model or a single tool, so how do you keep that stack maintainable when providers, models and tools keep changing every few weeks. Moonshot AI’s Kosong targets this problem as an LLM abstraction layer for agent applications. Kosong unifies message structures, asynchronous tool orchestration and pluggable chat…
How do we teach AI agents to reliably find and click the exact on screen element we mean when we give them a simple instruction? A team of researchers from ML Foundations has introduced Gelato-30B-A3B, a state of the art grounding model for graphical user interfaces that is designed to plug into computer use agents…
Kosmos, built by Edison Scientific, is an autonomous discovery system that runs long research campaigns on a single goal. Given a dataset and an open ended natural language objective, it performs repeated cycles of data analysis, literature search, and hypothesis generation, then synthesizes the results into a fully cited scientific report. A typical run lasts…
Reliable multi-agent systems are mostly a memory design problem. Once agents call tools, collaborate, and run long workflows, you need explicit mechanisms for what gets stored, how it is retrieved, and how the system behaves when memory is wrong or missing. This article compares 6 memory system patterns commonly used in agent stacks, grouped into…
In this tutorial, we explore how neural memory agents can learn continuously without forgetting past experiences. We design a memory-augmented neural network that integrates a Differentiable Neural Computer (DNC) with experience replay and meta-learning to adapt quickly to new tasks while retaining prior knowledge. By implementing this approach in PyTorch, we demonstrate how content-based memory…
Every time you prompt an LLM, it doesn’t generate a complete answer all at once — it builds the response one word (or token) at a time. At each step, the model predicts the probability of what the next token could be based on everything written so far. But knowing probabilities alone isn’t enough —…
