Mosaic is a framework dedicated to the comparison of AI models.
Project description
#MOSAIC Models for AI comparator
Mosaic is a framework dedicated to the comparison of AI models. It is often very difficult to choose the best AI model for a specific problematic and multiple o ptions are available, including choices over the model hyper-parameters. It is t empting to try different options and to compare them to get the best performance /resource ratio. But this kind of test can be pretty time consuming from the dev elopper point of vue. The Mosaic framework eases the automation of the program g eneration and provides tools to help the study (Database, plot system…)
Mosaic is a python framework based on Pytorch. From a simple configuration file, a set of pipelines is generated including all the steps of the data treatment ( data loading, formatting, normalization, post-treatment…) and the model training itself. The framework executes all these pipelines in a parallel way and store all the results in a database and in differents files. Some facility are offered to pause/resume and monitor the run. A plot module helps getting some compact a nd graphical representations to ease the interpretation of this data.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for ml_mosaic-0.0.4-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | acdf4f441484229b5aeb2db53b877d7e4068c196325519c9a2b0fca6f6ddbf78 |
|
MD5 | 4fbafc503501fd58d9ade059463b2fe2 |
|
BLAKE2b-256 | 984406dacb72f9eaa9f31c585e947492c69fac5d5a505f0ff3759b829debc2a5 |