What Surrounds Us takes its title literally. You play as a circle surrounding a hole in its middle - it looks like a donut with frosting. You work together with other sentient moving circles, sometimes helping them joyfully meet with others. And you traverse a large map that's composed entirely of, you guessed it, a […]

Reuters reports that the OpenAI agent that hacked Hugging Face had been free for a week before the company noticed.

Friendly hobby machine or serious production tool? Here’s how to know which one is for you.

Museums are embracing data-driven curation and a shifting technology landscape.

Plus, email as turn-based combat (which is really just regular email).

These e-readers are more useful in their old age than you might think.

Researchers say TikTok, X, and Meta aren't providing data they're legally required to.

I packed these Leatherology totes with laptops, chargers, and everything else my workday demands. Months later, they’re still the bags I keep reaching for.

Safe AI chat for kids Discussion | Link

Warner Bros. Discovery has sued Amazon for 'inducing' its employees to breach their contracts.

Over the last several years, the Imax format has taken on a few different forms—but only a few dozen theaters are capable of screening it in its full glory.

I Run Bare-Metal Kubernetes on $200 of Scrap Hardware (And Why I Burned 3 SD Cards...

Tokenization → embeddings → causal Transformer → LM head → softmax → loss → backpropagation. No TensorFlow, no PyTorch, and no hidden autograd. Repository: tiny-language-model-neuro-js . Most explanations of language models present correct formulas but hide the path between them inside a framework. I wanted the opposite: one small scenario where every scalar is visible and where the terminal clearly shows incorrect answers before learning and correct answers after it. The project now has one command: node src/train.js --generalize --adaptive-teach It requires Node.js 18.19+ and has no dependencies. The result first The model is queried immediately after random initialization: BEFORE TRAINING — random, usually wrong answers > can human read ? model: ? <unk> ... expected: human can read. [WRONG] > can fish swim ? model: ? <unk> ... expected: fish can swim. [WRONG] > can cat read ? model: ? <unk> ... expected: cat cannot read. [WRONG] After pre-training, SFT, and adaptive SFT, the same model produces: FINAL ANSWERS AFTER ADAPTIVE SFT > can human read ? model: human can read. [CORRECT] > can fish swim ? model: fish can swim. [CORRECT] > can bird fly ? model: bird can fly. [CORRECT] > can cat read ? model: cat cannot read. [CORRECT] Rehearsal controls preserved: 14/14. Stable criterion reached 11 times in a row. The initial text varies because initialization is random. The final acceptance criterion does not: all answers must be correct, every target token must have at least 95% probability, and the complete check must pass more than ten times consecutively. What remains after removing the extra modes The code previously contained several debug and training modes. They were useful while experimenting but obscured the main idea. The final version keeps one educational pipeline: text → word tokenization → token IDs → token + position embeddings → two causal Transformer blocks → multi-head self-attention → two-hidden-layer FFN → LM head → softmax → next-token probabilities