> get info N.N.N.
> Latest Version: V.2.4-m
> Purpose: Built not out of a need for another library, but as a byproduct of engineering curiosity taken too far.
N.N.N. is a bottom-up, manual reconstruction of deep learning mechanics, built from scratch to see exactly how the wheels turn under the hood.
(Just because we can)
> Version tags: -T (Language Model); -m, -M (Neural Network Library)
> Link to github: here
> get versioninfo all
> Warning! It is highly recommended to read the documentation before use as the information provided here is not sufficient for total understanding.
> Version: V.2.4-m
> Date Released: 2026-04-06
> Information: Added the option to set random seed and Early Stopping. Fixed Validation Error not working as expected when using Min-Max or Zscore.
> Documentation: here
> Link to github: here
> Version: V.2.3.1-m
> Date Released: 2026-03-01
> Information: Bug fix, Fixed He (Kaiming) Initialization
> Documentation: Use the documentation for V.2.3-m
> Link to github: here
> Version: V.2.3-m
> Date Released: 2026-01-04
> Information: Added Batch Normalization, thus improving the stability in larger models.
> Documentation: here
> Link to github: here
> Version: V.2.2-m
> Date Released: 2025-12-30
> Information: Added Vectorized Batching and small optimizations in code.
> Documentation: Use the documentation for V.2.1-m (ideally) or V.2.0-m
> Link to github: here
> Version: V.2.1-m
> Date Released: 2025-12-25
> Information: Added automatic dataset shuffling and validation checkpointing.
> Documentation: here
> Link to github: here
> Version: V.2.0-m
> Date Released: 2025-12-7
> Information: Added classification with an update to the structure and usage of the library.
> Documentation: here
> Link to github: here
> Version: V.1.0-m
> Date Released: 2025-10-31
> Information: Input regularization (Min-Max, Zscore) and more activation functions (No activation, ReLU) added.
> Documentation: here
> Link to github: here
> Version: V.0.2-M
> Date Released: 2025-09-21
> Information: Dropping .exe files in favor of pure .py files. ADAM optimization and Loss regularization (L1, L2) added.
> Documentation: Not available
> Link to github: here
> Version: V.0.2-m
> Date Released: 2025-05-13
> Information: The first version with backpropagation. Model size can be set.
> Documentation: Not available
> Link to github: here
> Version: V.0.1-T
> Date Released: 2025-04-01
> Information: Model size can be set and uses Numerical Differentiation in C++.
> Documentation: here
> Link to github: here
> Version: V.0.1-t
> Date Released: 2025-03-01
> Information: A fixed model size using Numerical Differentiation.
> Documentation: here
> Link to github: here
> Version: V.0.1-beta
> Date Released: 2024-12-21
> Information: A fixed model size using Pytorch as the engine.
> Documentation: Not Available
> Link to github: here
> Version: V.0.1-alpha
> Date Released: 2024-11-18
> Information: A fixed model size with gradient formulas calculated manually (Manual Differentiation).
> Documentation: here
> Link to github: here
> Version: Pre-Alpha V.0.1
> Date Released: 2024-11-01
> Information: The first version of N.N.N. It relies on random parameter changes to "learn".
> Documentation: Not Available
> Link to github: here
> last_update webpage
> Last update: 2026-06-27
> return