Knowledge Atlas

81 concepts · 7 clusters · directed prerequisites

Mathematical Foundations8Classical Machine Learning10Deep Learning14Large Models17Systems and Compute16Alignment and Safety8Applications8

• Attention — 7
• Transformer — 7
• Supervised Learning — 6
• Pretraining — 6
• Instruction Tuning — 6
• Distributed Training — 6
• Inference Serving — 6
• Linear Algebra — 5
• Backpropagation — 5
• Embeddings — 5
• Agents — 5
• RLHF — 5
• Probability — 4
• Optimization — 4
• Perceptron — 4
• Diffusion Models — 4
• KV Cache — 4
• Retrieval Augmentation — 4
• Long Context — 4
• Multimodality — 4
• GPU Architecture — 4
• Memory Bandwidth — 4
• FlashAttention — 4
• Hallucination — 4
• Evaluation Benchmarks — 4
• Information Theory — 3
• Unsupervised Learning — 3
• Reinforcement Learning — 3
• Dimensionality Reduction — 3
• Generalization and Overfitting — 3
• Normalization — 3
• Convolutional Networks — 3
• Recurrent Networks — 3
• Quantization — 3
• Tool Calling — 3
• CUDA — 3
• Tensor Cores — 3
• Mixed Precision — 3
• Vector Databases — 3
• Interpretability — 3
• Code Generation — 3
• Search and Recommendation — 3
• Machine Translation — 3
• Computer Vision — 3
• Calculus — 2
• Graph Theory — 2
• Statistical Inference — 2
• Numerical Computing — 2
• Decision Trees — 2
• Ensemble Learning — 2
• Activation Functions — 2
• Residual Connections — 2
• Self-Supervised Learning — 2
• Generative Adversarial Networks — 2
• Graph Neural Networks — 2
• Positional Encoding — 2
• Scaling Laws — 2
• Context Window — 2
• LoRA — 2
• Mixture of Experts — 2
• Chain of Thought — 2
• Operator Fusion — 2
• Data Parallelism — 2
• Batch Scheduling — 2
• Compiler Optimization — 2
• DPO — 2
• Red Teaming — 2
• Jailbreaks and Defense — 2
• Protein Structure — 2
• Autonomous Driving — 2
• Scientific Computing — 2
• Support Vector Machines — 1
• Clustering — 1
• Feature Engineering — 1
• Tokenization — 1
• Speculative Decoding — 1
• Tensor Parallelism — 1
• Pipeline Parallelism — 1
• ZeRO — 1
• Reward Models — 1
• Speech Recognition — 1

• 81 — Visible nodes
• 121 — Visible edges
• 3.0 — Average degree
• 60 — FPS

Supervised Learning

Transformer

Backpropagation

Embeddings

Probability

Optimization

Distributed Training

Instruction Tuning

Agents

Inference Serving

FREE ORBIT

Nothing selected

AtlasSphereLayersSignal flowLabelsAuto-orbit−100%＋NightReset

Drag to rotate · wheel or +/− to zoom · click a node to focus
Shift+click a second node for shortest path · Esc to clear