Baloot

Small, focused Python helpers for experiments and day-to-day developer workflows. The package is pure Python, has no required runtime dependencies, and keeps the API flat so you can import most helpers directly from baloot.

Overview

• Lightweight utilities for seeding, file helpers, settings, Torch ergonomics, plotting, and small modules.
• Optional integrations: install the libraries you actually use (e.g., torch, numpy, matplotlib).
• A simple, readable API surface intended for quick reuse in scripts and research code.

Installation

pip install baloot

API Reference

Constants

import baloot

print(baloot.pi)
# tensor(3.1416)

print(baloot.e)
# tensor(2.7183)

print(baloot.sqrt2)
# tensor(1.4142)

print(baloot.sqrt3)
# tensor(1.7321)

print(baloot.sqrt5)
# tensor(2.2361)

print(baloot.ln2)
# tensor(0.6931)

print(baloot.ln10)
# tensor(2.3026)

print(baloot.c)
# tensor(299792458)
assert baloot.c == baloot.light

print(baloot.golden)
# tensor(1.6180)

Reproducibility

import baloot

assert baloot.seed == baloot.seed_everything

baloot.seed(1234)

Alternatively, you can seed different libraries separately.

import baloot

baloot.seed_python(1)
baloot.seed_numpy(2)
baloot.seed_torch(3)
baloot.seed_cuda(4)

File Management

Saving data:

import baloot

data = {
    "name": "John Doe",
    "age": 222
}

baloot.funnel("data.json", data)

Later on you can load the data:

import baloot

data = baloot.funnel("data.json")

Triangular

A few exotic triangular functions have been added to Baloot to complete already rich triangular API of PyTorch.

import baloot

baloot.arccot(baloot.pi * 3 / 4)
baloot.arccoth(baloot.pi * 3 / 4)
baloot.cot(baloot.pi * 3 / 4)
baloot.coth(baloot.pi * 3 / 4)
baloot.hacoversin(baloot.pi * 3 / 4)
baloot.haversin(baloot.pi * 3 / 4)

assert baloot.hacoversine == baloot.hacoversin
assert baloot.haversine == baloot.haversin

Tensor Operations

Anti-diagonal:

import torch
import baloot

matrix = torch.randn(10, 10)
baloot.antidiagonal(matrix)

Batched trace operation:

import torch
import baloot

matrices = torch.randn(10, 5, 5)
traces = baloot.trace(matrices)
assert traces.shape == (10, 1)

Copyright

MIT License. Copyright (c) 2025 Taha Shieenavaz. See LICENSE.