Tools & simulators

• Tokenizer Playground — Type anything and watch a real BPE tokenizer split it — tokens, ids, and counts, computed in your browser.
• Attention Visualizer — Step through a sentence head by head and see attention as heatmaps and arcs.
• Sampling Lab — Turn the temperature, top-k, and top-p knobs on a toy distribution and watch the output odds change.
• KV Cache & VRAM Calculator — Layers, heads, precision, context, batch — the memory formulas on screen, with your numbers in them.
• GPU Sizing Calculator — Weights plus KV cache plus overhead, by precision — mapped to instance classes on AWS, Azure, and GCP.
• Token Cost Estimator — Estimate monthly spend from dated, sourced price tables — never from memory.
• Prompt vs RAG vs Fine-tune — A decision tree that shows its reasoning at every step and ends in a defensible recommendation.
• Embedding Space Explorer — A 2D projection of real embeddings — hover, cluster, and see meaning as geometry.
• Scaling Law Plotter — Loss versus params, data, and compute from the published scaling-law formulas, cited on screen.
• Cloud Model Selection Matrix — Filterable comparison of managed LLM services across AWS, Azure, and GCP — downloadable as CSV.