50 lines (37 loc) · 1.56 KB

Changelog

[0.0.1] - 2024-12-28

First logged version of the library ✅
Tensor class in Tensor module created with: full Tensor functionalities, error handling and notebook tests + pytorch ui comparisons🚀
dtype Enum class in Tensor module created, involving 2 datatypes for now: int64 and float64
Primary documentation of the steps added in tests/ 🧪
Primary repository structure design to include: requirements.txt, LICENCE.md, VERSION and docs/

[0.0.2] - 2024-12-30

Allowed numpy ndarray input to tensor through the static method validating the data input
Allowed direct conversion between tensor and numpy

[0.0.3]

Added prototype 1: scratch of all classes

[0.0.4]

Added prototype 2: enhancement of all classes with first functional outcome

[0.0.5]

Branched to prototype 3 to merge the extension classes (dataset, laoder and transform) with all functional modules

[0.0.6]

Updated Linear, Parameter, Module to include abstracts and other functionalities

[0.0.7]

Added activation.py modularizing its usage as in pytorch

[0.0.8]

Fixed peformance

[0.0.9] - 2025-01-05

Updated UML + report
Made backward() a decorator
Added synthetic data tests
Finalizing project

[1.0.0] - 2025-01-08

First official functional version of the neural network library 🔥
Final report, presentation and defense day ✔️
Library structured and structured
Added setup.py and tests

[1.0.1] - 2025-02-11

made it pip installable
added test example for users
reorganized the repo structure
changed libname to che3le