HashSlap-Summer-of-Code/ml-core

Design a Unified ML Experiment Tracking Framework

Ouverte

#7 ouverte le 18 juin 2025

 (0 commentaire) (0 réaction) (0 personne assignée)Jupyter Notebook (11 forks)auto 404
Intermediateenhancementgood first issuehacktoberfesthssoc

Métriques du dépôt

Stars
 (3 étoiles)
Métriques de merge PR
 (Métriques PR en attente)

Description

Description:
Develop a lightweight and consistent way to log and visualize machine learning experiment results across all subfolders (e.g., neural-networks, supervised-learning, anomaly-detection). This helps contributors compare results over time and improve reproducibility.

Expected Tasks:

  • Create a Python utility (e.g., experiment_logger.py) that logs metrics like accuracy, loss, and hyperparameters to a .csv file.
  • Add basic plotting functionality using matplotlib or seaborn.
  • Place the utility in a new folder like utils/ or tools/.
  • Create a sample log for an existing implementation (e.g., Perceptron).
  • Write a README.md in the root directory explaining:
    • How to use the logger.
    • Required libraries.
    • How to integrate it into a new or existing ML script.

Stretch Goal:

  • Explore integration with lightweight experiment trackers like MLflow or Weights & Biases, while keeping setup minimal.

Guide contributeur