MLBoardKit

A Python library that provides utilities for streamlined data processing, model training, and analysis tasks in machine learning workflows.

mlboardkit offers easy CLI commands and Python interfaces for dataset quality checks, format conversion, metric computation, plot generation, and model training — with support for popular frameworks and minimal setup.

Install

# from source (editable)
pip install -e .

# from PyPI (published)
pip install mlboardkit

Quick start

# After installing mlboardkit, import via the mlboardkit namespace
from mlboardkit.data_utils.dataset_processor import main as dataset_processor_main
from mlboardkit.analysis_tools.metrics_utils import classification_report

report = classification_report([1,0,1], [1,0,0])

CLI via python -m:

python -m mlboardkit.data_utils.dataset_processor quality-check dataset.csv --report report.json
python -m mlboardkit.data_utils.data_converter convert input.json output.csv --format csv
python -m mlboardkit.analysis_tools.plot_metrics training_log.json --plot-type training --output curves.png
python -m mlboardkit.model_utils.train_model --model-name bert-base-uncased --train-file train.jsonl --epochs 3

Python requirement: 3.9+

Full usage and CLI examples are in usage.md. Here is a demo notebook that demonstrates the usage of this library in a ML project.