AI Paper Atlas

Research map / Probabilistic and scientific modeling / Dynamical systems

Model Zoos: A Dataset of Diverse Populations of Neural Network Models

Konstantin Schürholt, Diyar Taskiran, Boris Knyazev, Xavier Giró-i-Nieto, Damian Borth

NeurIPS 2022 · Probabilistic and scientific modeling › Dynamical systems · approach: Latent models of neural activity · 4 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Vary hyperparameters instead of varying only the seedCNN (s) model zoos · MNISTaccuracy diversity (standard deviation): 0.9 -> 30.7Improved
Fix rather than randomly vary the seeds while varying hyperparametersHyp-10 model zoos · MNISTaccuracy distribution: 79.9 (30.7) -> 80.3 (30.3)No clear effect
Use layer-wise weight statistics instead of raw weights as prediction featuresLinear model property prediction · MNIST, CNN (s), Seedaccuracy R² (%): 92.3 -> 98.7Improved
Use layer-wise weight statistics instead of raw weights as prediction featuresLinear model property prediction · MNIST, CNN (s), Hyp-10-raccuracy R² (%): -11.2 -> 69.4Improved
Use layer-wise weight statistics instead of raw weights as prediction featuresLinear model property prediction · CIFAR10, CNN (s), Seedaccuracy R² (%): 75.3 -> 96.0Improved
Use fixed rather than varied seeds when varying hyperparametersCNN (s) model zoos · MNISTaccuracy prediction R² (%) from raw weights: -11.2 -> 66.5Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy