dselsam/certigrad
Bug-free machine learning on stochastic computation graphs
This project helps machine learning researchers and practitioners build complex AI models with absolute certainty about their mathematical correctness. It takes your description of a stochastic computation graph and verifies that the gradient calculations are provably correct, producing a debugged and mathematically sound model. This is for machine learning researchers, deep learning engineers, or data scientists working on models where correctness is paramount.
399 stars. No commits in the last 6 months.
Use this if you need to develop machine learning systems, especially those with random elements, and you want to mathematically prove their core gradient computations are correct, rather than relying solely on empirical testing.
Not ideal if you are looking for a general-purpose machine learning framework for rapid prototyping or if you prioritize ease of development and flexibility over formal mathematical guarantees.
Stars
399
Forks
36
Language
Lean
License
Apache-2.0
Category
Last pushed
Mar 03, 2019
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/dselsam/certigrad"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
jolars/sortedl1
Python package for Sorted L-One Penalized Estimation (SLOPE)
gugarosa/opytimizer
🐦 Opytimizer is a Python library consisting of meta-heuristic optimization algorithms.
SENATOROVAI/gradient-descent-sgd-solver-course
Stochastic Gradient Descent (SGD) is an optimization algorithm that updates model parameters...
hiroyuki-kasai/SGDLibrary
MATLAB/Octave library for stochastic optimization algorithms: Version 1.0.20
softmin/ReHLine-python
Regularized Composite ReLU-ReHU Loss Minimization with Linear Computation and Linear Convergence