PaperMap turns landmark AI research papers into visual, interactive learning experiences. Explore the Transformer, GPT-3, InstructGPT, RAG, and LoRA — rebuilt as production-grade explainers.
Each paper is a self-contained interactive guide with visual demos, architecture breakdowns, quizzes, and research-accurate detail from the original publications.
The paper that started it all. Interactive walkthrough of the Transformer architecture with tokenization, embeddings, self-attention, multi-head attention, positional encoding, and full encoder-decoder visualization.
Open Explainer →How 175 billion parameters unlocked in-context learning. Interactive few-shot demos, scaling law visualizations, benchmark tables, data contamination analysis, and the full GPT-3 model family.
Open Explainer →The RLHF paper that paved the way for ChatGPT. Interactive 3-step pipeline (SFT → Reward Model → PPO), human evaluation results, bias analysis, and qualitative examples.
Open Explainer →An interactive deep dive into the original RAG paper: dense retrieval with DPR, BART generation, non-parametric memory with Wikipedia, and benchmark gains on knowledge-intensive QA.
Open Explainer →An interactive walkthrough of LoRA covering low-rank updates, parameter-efficient fine-tuning, practical deployment tradeoffs, and why LoRA matches full fine-tuning at a fraction of the cost.
Open Explainer →A scalable, static-first platform that transforms dense research papers into interactive learning products — no frameworks, no build tools, no dependencies.
Every number, formula, and architectural detail is sourced directly from the original paper. Section references included throughout.
Tokenizers, attention maps, few-shot playgrounds, scaling charts, and quiz blocks — learn by building intuition, not just reading.
Responsive layouts, touch targets, and reduced-motion support. Works perfectly on phones, tablets, and desktops.
Pure HTML, CSS, and vanilla JavaScript. Deploy anywhere — GitHub Pages, Netlify, Vercel, or any static host.
Semantic HTML, ARIA labels, keyboard navigation, skip links, focus styles, and prefers-reduced-motion support.
Shared typography, color palette, components, and interaction patterns across all paper explainers.
PaperMap grows one paper at a time. Each explainer is handcrafted with the same production-quality standards.
Transformer architecture, tokenization, embeddings, self-attention, multi-head attention, positional encoding, training.
In-context learning, scaling laws, 175B architecture, benchmark results, data contamination, societal impact.
3-step RLHF pipeline, reward modeling, PPO, human evaluations, alignment tax, bias analysis.
Dense retrieval with DPR, BART generation, non-parametric memory, and SOTA gains on knowledge-intensive QA.
Low-rank updates for efficient LLM fine-tuning with massive parameter savings and zero extra inference latency.
BERT, diffusion models, multimodal systems, and more landmark papers coming soon.