RESEARCH IS OPEN
Rudrite Research is live
Interactive, animated explainers of the AI & ML papers that matter — now open at research.rudrite.com.
The workshop is no longer quiet. Rudrite Research is open — the first wing of Rudrite, and the one we kept hinting at.
Research turns landmark AI and ML papers into interactive, animated, deeply-read explainers. Not summaries. The whole paper, made legible: every contribution explained in order, with bespoke exhibits you can play, computed from the paper's real formulas — and every quote verbatim, checked against the source.
What's there now
Sixteen landmark papers, rebuilt end to end:
- Foundations — Attention Is All You Need, GPT-3, BERT, Chinchilla scaling laws, Mixtral, Mamba.
- Systems & kernels — FlashAttention, PagedAttention, Megatron-LM, ZeRO.
- Model architectures — DeepSeek-V3, Qwen3, OLMo 2, MiniMax-01, Gemma 4, and DeepSeek-R1.
Plus side-by-side comparisons — like Transformers vs Mamba, FlashAttention vs PagedAttention, and Dense vs Mixture-of-Experts.
The bar
Every explainer is played, then read: drag the words to see what attention looks at, slide the sequence length to watch a quadratic cost diverge from a linear one, toggle an MoE router to fire the experts. The exhibits are computed from the actual math, not stock diagrams. Quotes are the paper's own words. It's monochrome, fast, and free — no sign-up, no ads.
What's next
Research keeps growing — more papers, more comparisons, deeper exhibits. And the second wing, Academy, is in the works.
Start with Attention, or browse the whole library.