Interactive activities
Visual labs
Change assumptions, compare outcomes, and practice system decisions. Every simulation is deterministic and discloses its simplifications.
23
labs & sandboxes
WebAssembly AI Runtime & Micrograd Sandbox
Write and execute Python in WebAssembly (Pyodide) with live SVG autograd DAG visualization, FlashAttention online softmax, and KV cache page indexing.
FlashAttention-3 SRAM Tiling & Online Softmax
Visualize HBM vs. on-chip SRAM tiling and step through row-level online softmax rescaling without materializing O(N²) attention.
MoE Dynamic Routing & Expert Parallelism
Simulate Top-2 gating softmax, capacity factor C, dynamic bias adaptation, and live node failure chaos across 16–64 experts.
PagedAttention Virtual Memory Simulator
Explore non-contiguous physical GPU VRAM block tables, prefix cache sharing, and fragmentation elimination.
Speculative Decoding & Rejection Sampling
Model γ-token speculation, rejection sampling α_i = min(1, p/q), residual distribution correction p'(x), and HBM bandwidth savings.
GRPO vs. PPO Reasoning Alignment
Explore critic-free group-relative advantage normalization, Schulman k3 non-negative KL penalties, zero-variance guards, and 50% VRAM reductions.
Tokenization and Context Window Visualizer
Inspect deterministic token approximations and allocate a finite context budget.
Sampling and Decoding Explorer
Explore how temperature and top-p reshape a fixed token distribution.
Prompt RAG Fine-Tuning or Tool-Use Decision Lab
Match system symptoms and constraints to an appropriate intervention strategy.
RAG Pipeline and Retrieval Parameter Visualizer
Tune chunk size retrieval breadth hybrid weight and reranking in a deterministic pipeline.
LLM Cost Latency and Reliability Simulator
Model request economics latency percentiles and fallback effects under disclosed assumptions.
Evaluation Strategy Designer
Assemble a balanced evaluation plan from risks cases metrics review and release gates.
Streaming JSON Repair & Buffer Playground
Visualise incomplete stream buffering, UTF-8 sequence cuts, and token-repair delimiters.
OpenAI Codex Codebase & Control Path Visualizer
Trace interactive request logic, TUI UI steps, MCP integrations, and sandbox execution chains.
Embedding Space Explorer
Inspect deterministic vector geometry similarity measures and ranking changes across a small semantic corpus.
Chunking and Reranking Laboratory
Compare boundary-aware chunks candidate breadth reranker strength context cost and retrieval quality under explicit assumptions.
Agent Loop and Tool Selection Simulator
Step through bounded agent states and select the correct action tool approval or stop condition for each observation.
Evaluation Metric Comparison Lab
Match failure modes to deterministic metrics model graders human review and production measures without collapsing quality into one score.
Inference Batching and Queueing Simulator
Model deterministic arrival rate batch size service time cache pressure throughput utilization and tail-latency trade-offs.
Prompt-Injection Threat-Model Exercise
Classify direct and indirect injection paths then assemble independent authorization isolation validation and response controls.
Account Abstraction & Policy Engine Lab
Simulate ERC-4337 smart account execution, Paymaster gas sponsorship, session keys, and Policy Engine security rules (allowlists, spending caps, HITL gates) under normal and adversarial prompt-injection workloads.
FlashAttention-3 SRAM Tiling & Online Softmax Lab
Simulate GPU memory hierarchy, SRAM tiling, and online softmax rescaling to understand O(N) memory footprints and zero intermediate HBM traffic.
Mixture-of-Experts Dynamic Routing & Expert Parallelism Lab
Simulate top-2 gating softmax over fine-grained expert pools, token capacity factors, auxiliary load balancing loss, and chaos skew injections.