ModLens · The CAS Architecture Explorer
See how intelligent
systems are built.
ModLens is the interactive CAS index of modern AI architectures. Every diagram is a lens: focus a component and read what it actually does — data flow, mathematics, strengths and limits.
Convolutional Neural Network
A network built from convolutional layers that exploit the spatial structure of images: local receptive fields, shared weights and hierarchical feature composition.
5 components · open the lens →
ResNet
Residual networks make depth trainable by letting layers learn corrections to the identity: each block computes F(x) + x instead of an unreferenced transform.
4 components · open the lens →
EfficientNet
A family of vision networks derived from a compound scaling rule that balances depth, width and input resolution under a fixed compute budget.
4 components · open the lens →
Vision Transformer
Images treated as sequences of patches processed by a transformer: no convolutions, spatial relationships learned through attention and position embeddings.
5 components · open the lens →
YOLO
You Only Look Once: object detection formulated as a single regression problem over a grid, enabling real-time detection in one network pass.
5 components · open the lens →
R-CNN Family
Region-based detectors: propose candidate objects, then classify and refine each proposal. Faster R-CNN made proposal generation part of the network itself.
5 components · open the lens →
Transformer
The architecture that replaced recurrence with attention: parallelizable sequence modeling whose scaled descendants power modern foundation models.
5 components · open the lens →
Encoder–Decoder
Two coupled transformers: an encoder reads the full input bidirectionally; a decoder generates the output conditioned on that reading.
5 components · open the lens →
GPT-style Decoder
A stack of causal transformer blocks trained to predict the next token. At scale, this single objective yields general-purpose language capability.
4 components · open the lens →
Mixture of Experts
Scaling capacity without scaling compute per token: feed-forward layers are replaced by many expert subnetworks, with a router selecting a few per token.
6 components · open the lens →
Multimodal Architecture
Encoders for different modalities — vision, language, audio — connected through shared representation spaces or a shared decoder.
5 components · open the lens →
Agent Architecture
A model embedded in a loop: it observes state, plans, calls tools, reads results and revises — with memory and supervision as first-class components.
7 components · open the lens →
Memory Architecture
The layered state of an intelligent system: context windows, vector stores, structured records and episodic traces — engineered as one memory hierarchy.
5 components · open the lens →
World Models
Models that learn predictive representations of environments: compressing observations into latent states and modeling their dynamics to imagine futures.
5 components · open the lens →
