scitex-repro 0.1.7

Reproducibility utilities (random state, timestamps, hashing)

scitex-repro

Reproducibility utilities — RNG seeding, ID generation, timestamps, array hashing.

Full Documentation · uv pip install scitex-repro[all]

---

Problem and Solution

#	Problem	Solution
1	"Seed everything" requires 5+ lines of boilerplate — random.seed() + np.random.seed() + torch.manual_seed() + torch.cuda.manual_seed_all() + tf.random.set_seed() + os.environ[PYTHONHASHSEED]	RandomStateManager(seed=42) — one call seeds every framework detected in the env; .reset() rewinds mid-experiment
2	Experiment run directories collide — two parallel runs overwrite each other	gen_ID() + hash_array() — unique directory names like 20260423_2155_abc12345; deterministic array fingerprints for integrity checks

Installation

pip install scitex-repro

Architecture

src/scitex_repro/
├── __init__.py              # public re-exports
├── _RandomStateManager.py   # cross-framework RNG seeding (random / numpy / torch / tf)
├── _gen_ID.py               # 20260423_2155_<hash> directory IDs
├── _gen_timestamp.py        # filesystem-safe timestamps
├── _hash_array.py           # deterministic NumPy / pandas fingerprints
└── _config.py               # env-var + config defaults

1 Interfaces

Python API

from scitex_repro import RandomStateManager, gen_ID, hash_array

# Cross-framework RNG seeding
rng = RandomStateManager(seed=42)
data = rng("data").random(100)
rng.verify(data, "my_data")
rng.reset()

# Deterministic IDs and array fingerprints
exp_id = gen_ID()                         # "20260423_2155_abc12345"
fingerprint = hash_array(data)

Demo

flowchart LR
    seed["seed=42"] --> rsm["RandomStateManager"]
    rsm --> py["random"]
    rsm --> np["numpy"]
    rsm --> torch["torch (cpu+cuda)"]
    rsm --> tf["tensorflow"]
    rsm --> hashenv["PYTHONHASHSEED"]
    rsm --> data["rng('data').random(100)"]
    data --> fp["hash_array(data)"]
    rsm --> id["gen_ID()"]
    id --> dir[("20260423_2155_abc12345/")]
    fp --> integrity[("array fingerprint")]

Quick Start

from scitex_repro import RandomStateManager, gen_ID, hash_array

rng = RandomStateManager(seed=42)
data = rng("data").random(100)
print(gen_ID(), hash_array(data))

Part of SciTeX

scitex-repro is part of SciTeX. Install via the umbrella with pip install scitex[repro] to use as scitex.repro (Python) or scitex repro ... (CLI).

Four Freedoms for Research

0. The freedom to run your research anywhere — your machine, your terms.
1. The freedom to study how every step works — from raw data to final manuscript.
2. The freedom to redistribute your workflows, not just your papers.
3. The freedom to modify any module and share improvements with the community.

AGPL-3.0 — because we believe research infrastructure deserves the same freedoms as the software it runs on.

License

AGPL-3.0-only (see LICENSE).

---