Beneath the interface · Taking shape now
Dont0'tt1 bet2 boredt3.t4 Gett5 lt6oredt7.t8
Five words, one apostrophe, two full stops. Flip the switch for a glimpse beneath the interface.
We're building visual, deeply satisfying explanations of LLMs and the technology shaping your world—without hand-waving or unexplained jargon.
Why this site is being built
The internet gives you answers.
We're building the lore.
Lored Not Bored will go beneath the prompt—from tokens and embeddings to attention, training, hallucinations, and the important steps people usually skip.
The standard we're building toward
Three rules from day one.
- R1No hand-wavingIf an explanation needs a skipped step, we'll say so plainly.
- R2No unexplained jargonTechnical words will come with useful, human definitions.
- R3Show it, then say itWe'll make the idea visible before asking the words to carry it.
A preview of the approach
Tap the dotted words
A reads your prompt, turns it into , and then one token at a time.
transformer — the architecture behind many modern language models. Its key move lets tokens weigh the relevance of other tokens in context.
logits — raw scores for possible next tokens. They are converted into a probability distribution before a choice is made.
samples — selects a next token from that distribution. The settings control whether the result is more predictable or more varied.
3 terms · 0 assumed
Still beneath the surface
The lore is loading.
We're still drawing, testing, and fitting the pieces together. Lored Not Bored isn't live yet, but it is on the way.
Coming soon