AntixK/PyTorch-Model-Compare
Compare neural networks by their feature similarity
This tool helps machine learning researchers and practitioners understand how different neural network models learn and represent data. It takes two PyTorch neural networks and optionally a dataset as input. It then generates a numerical similarity score or a heatmap that visually compares how similar the internal feature representations of these networks are, layer by layer or across different datasets. This is useful for anyone developing or evaluating deep learning models.
379 stars. No commits in the last 6 months.
Use this if you need to compare how two neural networks process information internally, evaluate the impact of architectural changes, or understand model behavior across different datasets without just looking at final performance metrics.
Not ideal if you are looking for tools to compare model performance solely based on accuracy or other output metrics, or if you are not working with PyTorch neural networks.
Stars
379
Forks
41
Language
Python
License
MIT
Category
Last pushed
May 17, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/AntixK/PyTorch-Model-Compare"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
VectorInstitute/odyssey
A toolkit for developing foundation models using Electronic Health Record (EHR) data.
ycq091044/BIOT
BIOT - A framework for pretraining biosignals at scale. Large EEG pre-trained models.
woodRock/fishy-business
Machine Learning for Rapid Evaporative Ionization Mass Spectrometry for Marine Biomass Analysis...
soda-inria/carte
Repository for CARTE: Context-Aware Representation of Table Entries
harryjdavies/HeartGPT
Interpretable Pre-Trained Transformers for Heart Time-Series Data