N Neurarch Architectures Checks Data Docs Open the app

Resources / Datasets

Neurarch architecture corpus

36 neural network architectures kept as typed graphs rather than diagrams. Each entry carries every layer's type and parameters, the tensor shape propagated through it, an estimated parameter count, the verdict of 41 structural checks, and a link to the model.json an agent can fetch, edit and submit back for verification. Families span vision, language, recommendation, diffusion, biosignal and speech.

36 architectures CC0-1.0 Free to use Browse the architectures On Hugging Face

Get it

Index of every architectureneurarch.com/a/index.json
application/json
Hugging Face mirror (dataset viewer, load_dataset)huggingface.co/datasets/neurarch-ai/neurarch-architectures
text/html
 
curl -s https://neurarch.com/a/index.json | jq '.architectures[] | {key, params, verdict}'

What is in a row

layer count estimated parameter count input shape output shape verifier verdict

What this dataset is not

Every graph here passes our own verifier by construction: a template with a blocking finding is held back from the gallery rather than published. That makes this a corpus of designs that are structurally sound, not a sample of designs people actually write, and the wrong set to measure a checker's recall against. The grounding study is the one with failures in it.

Licence and citation

Released under CC0 1.0 Universal. Cite it as:

Neurarch. Neurarch architecture corpus. https://neurarch.com/d/architectures.html

The rest of the set

Neurarch structural check catalogue
The 41 structural checks Neurarch runs on a model graph, as data: id, severity, category, the condition that triggers it, why it costs something, and the fix.
41 checks · CC-BY-4.0
Verifier grounding study (264 graphs)
Clean reference architectures plus systematically corrupted variants (broken attention head divisibility, linear width mismatches, severed connections), each built as a real PyTorch model and run on a GPU.
264 graphs · MIT
Arch-Bench arena results
Frontier language models scored on design-from-spec tasks by a deterministic verifier rather than a human or an LLM judge.
18 model-split results · CC-BY-4.0
Arch-Bench task set
The task definitions behind the benchmark: design-from-spec and repair-and-extend instances for agents that build neural network architectures.
12 curated tasks, 8 fixtures · MIT
arch-design-sft: verified architecture-design SFT data
Supervised fine-tuning data for neural architecture design treated as structured graph editing.
3,010 verified examples · MIT
Structure, verdict and trained outcome triples
The corpus that pairs what a design looks like with what it did.
80 trained graphs · MIT
Verified architecture-design reasoning traces (Claude)
Spec to reasoning to design triples where the design is re-graded by the same deterministic verifier the benchmark uses, and only passing traces are kept.
306 verified traces · MIT
Verified architecture-design reasoning traces (Grok)
The same verified spec to reasoning to design triples as the Claude split, rejection-sampled from a different frontier model, so the two can be pooled for volume or held apart to see how much of the reasoning style is model-specific.
376 verified traces · MIT