Table tennis in the browser. Your phone is the paddle. Open pingpad on a laptop, tablet or TV, scan the QR code and swing. No app, no account. Phone and screen communicate directly, no game server in between. #gamedev# #WebGames# #WebRTC#

Table cleaning Napkins, foldedWith each passing month, Lila's bubble tea adventure kept expanding. She launched a YouTube channel, where she uploaded videos of her exploring shops, sharing her opinions and suggestions with her audience. The channel rapidly attracted a big follow

Table salt is fine but sea salt gives you more: - For sodium purposes common table salt will suffice - Sea salt can contain dozens or more minerals, some of them quite valuable to your health - Others matter only in trace amounts - You will get few minerals, if any, from common table salt The pretty colors and the taste are the least of it. @hubermanlab

Tabular data is finally getting its own era of ready-to-use foundation models, the way images and text already have. NVIDIA just open-sourced a new library for it. Title: Structured Data Models (NVIDIA/structured-data-models) URL: 📝 Overview A GPU-native, PyTorch-compatible library bundling foundation models, tensor types, and data processors for tabular and relational data. ❓ Problem Solved Structured data lagged behind images and text when it comes to GPU-native infrastructure, often forcing teams to retrain task-specific models from scratch. 💡 Method & Approach It ships tabular models like TabICLv2, KumoTabular, and TabFM alongside KumoRelational for graph-structured data. The core `TableTensor` abstraction losslessly tensorizes raw data on GPU, and models support both in-context learning and cached fit-predict inference. 🌍 Use Cases Numerical, categorical, datetime, text, and relational data types can all flow through pre-trained models with minimal code. It requires Python 3.11+ and PyTorch 2.7+, and installs directly via pip. 📊 Results The project claims new accuracy-efficiency frontiers for tabular prediction, plus zero-shot capability across diverse structured datasets. #MachineLearning# #GPU#

Tabular foundation models have mostly just been predicting a single target variable. A new paradigm learns the relationships across an entire table instead. Title: LimiX-2: A Contextual Mechanism Network Towards General Structured-Data Intelligence URL: 📝 Overview Contextual Mechanism Networks (CMN) learn the full joint distribution p(x, y) across a table, not just the target y, letting a single model handle classification, regression, missing-value imputation, and causal discovery. ❗ Problem it solves Existing PFN-based methods specialize in conditional prediction of the target only, without explicitly modeling the dependency structure between variables. ⚙️ Methodology Cell-level embeddings feed a 24-layer dual-axis transformer, trained via "Context-Conditional Masked Modeling" that masks part of the query rows and jointly predicts both features and targets. It's pretrained at scale on synthetic data generated from structural causal models. 📊 Results On TabArena (51 datasets), it hits an Elo of 1935 vs. 1818 for the baseline; on BCCO (156 datasets), cumulative wins are 2.06x higher. It also took 1st place on F1 score for causal skeleton recovery on all 6 tested datasets, all while using 4x fewer parameters than TabFM. 🔬 Use cases Since the same model handles classification, regression, and causal discovery, it could replace much of a data analysis pipeline with a single foundation model. #TabularData# #MachineLearning#