What it does This repo is a living bibliography of deep learning architectures, organized as a hierarchical mindmap and rendered into a text README. It covers CNNs, RNNs, GANs, reinforcement learning, memory networks, and capsule nets, with arXiv links and dates for each paper. The README is auto-generated from a plain text source file; contributors edit the .txt, not the Markdown.

The interesting bit The project treats paper lineage as a public utility rather than a literature review. The Coggle mindmap visualizes how branches split—object detection from CNNs, attention from seq2seq, WGAN from GAN—making the pace of 2014–2017 deep learning feel almost geological.

Key highlights

• Hierarchical coverage: CNNs (with sub-branches for detection, segmentation, super-resolution), RNNs, generative models, RL, memory networks, capsule nets
• Each entry links to the original paper; dates range from LSTM (‘97) to Transformer and Capsule Net (‘17)
• Auto-generated README: edit Neural Net Arch Genealogy.txt, the rest is scripted
• Open to PRs for new architectures “worth paying attention to”
• Visual mindmap hosted on Coggle with embedded PNG fallback

Caveats

• Stopped updating around 2017; no transformers beyond the original “Attention Is All You Need,” no modern LLMs, diffusion models, or state-space models
• “Automatically generated” is stated twice but the generation script isn’t visible in the README
• Some arXiv links may have rotted; no CI or validation is mentioned

Verdict Useful if you need a quick historical map of mid-2010s deep learning or want a template for crowdsourced taxonomy. Skip it if you need current architectures; this is a time capsule, not a living reference.