Model Tools

Diagnostics, analysis, and visualization for ML model development.

Purpose

This package provides utilities used across model sub-projects:

• analysis — Chain-length analysis, metrics computation
• viz — Training curves, market plots (OHLCV, peaks, volume)
• io — Thin wrappers around homebrewlibra I/O

What is NOT here:

• Data preparation (sliding window, standardize, split) → data-prep
• Model-specific tensor loading → model-core

Installation

# From PyPI
pip install model-tools-ml

# Editable (local)
pip install -e .

# With tests
pip install -e ".[tests]"

# From GitHub
pip install git+https://github.com/Alex-Glebov/model-tools.git

Package Structure

model_tools/
├── analysis/      # Chain analysis, metrics
│   └── analyze_chains.py
├── viz/           # Visualization
│   ├── plot_metrics.py
│   ├── plot_price_peaks.py
│   └── market_plots.py
└── io/            # I/O wrappers (homebrewlibra)

Usage

import model_tools
print(model_tools.__version__)  # 0.1.3

# Analysis
from model_tools.analysis import analyze_all_chains
min_len, max_len, mean_len, all_lengths = analyze_all_chains(data_path)

# Visualization
from model_tools.viz import plot_training_history
plot_training_history(config_path)

Tests

pytest tests/ -v

Version

0.1.3 — defined in pyproject.toml.

Author

Alex & Claude python@iitsp.com.au