Hypercube LCN

Status: Python package 1.2.x. Weights are packed depth, axis, tap, vertex.

Definitions

Symbol / term	Meaning
LCN	The Python class wrapping Core + Training.
dim	Hypercube dimension. Valid range [4, 24].
N	Vertex count, N = 2ᵈⁱᵐ. Field length.
z_max	Depth count; constructor 0 means use dim.
gather_span	Lookback window width, 2–6.
weights	All trainable weights: depth, axis, tap, vertex.

This package is the Python surface for HypercubeLCN (import hypercube_lcn). Full API reference: docs/Python_SDK.md. C++ integration guide: docs/CPP_SDK.md. Project home: github.com/dliptak001/HypercubeLCN.

HypercubeLCN is a Locally Connected Network on a Boolean hypercube: a deep feedforward network whose connectivity is the cube's own edges and whose weights are trained. It is built from two core classes.

The Core class is the network. It owns the weights and runs the forward pass: one field in, one field out, with a stack of intermediate fields written on the same cube in between.

The Training class walks that same pass in reverse. It accumulates gradients through every depth and steps the weights with Adam.

That is the whole architecture. There is no preprocessor, no reservoir, no separate readout — the cube is the model. In Python the two are wrapped by a single class, hypercube_lcn.LCN.

This is the opposite bet from the sibling projects. HypercubeEtalon, HypercubeWTF, and HypercubeCascade all put a frozen random hypercube stage in front of a small trained readout, and the point of those experiments is how far fixed dynamics can carry a thin classifier. This project asks instead: how much better does the hypercube do when every weight in it is trained?

HypercubeAI ecosystem

HypercubeCascade · HypercubeCNN · HypercubeESN · HypercubeEtalon · HypercubeHopfield · HypercubeLCN · HypercubeWorldModel · HypercubeWTF

📄 Foundational paper: Boolean Hypercubes as a Neural Substrate (D. C. Liptak, 2026)

HypercubeLCN is an experiment in the HypercubeAI project — our quest to systematically re-implement classical neural architectures on a Boolean hypercube topology instead of Euclidean grids or random graphs. The central thesis is "topology-native intelligence": the hypercube's algebraic structure (vertex-transitive symmetry, Hamming geometry, bitwise addressing) can serve as a first-class computational substrate.

A topology you don’t store — the graph is specified: connectivity is implicit in the vertex indices; with a seed and a few config scalars the whole reservoir reconstructs mathematically.
Perfect homogeneity — every vertex has the same degree and the same local world, so local dynamics mean the same thing everywhere — no structural favorites baked in by a random graph.
Cheap navigation — each neighbor is a few bit operations on the vertex index, not a pointer chase through a stored edge list, so walks stay arithmetic and cache-friendly.

Each product in the family is a different architecture on that same foundation.

The LCN

A locally connected network is wired like a convolutional layer — each unit reads only a small neighborhood — but where a CNN slides one shared kernel across every position, an LCN lets every position train its own private weights.

Here the hypercube supplies the neighborhoods. A vertex's neighbors are the indices one bit-flip away, and at every depth each vertex gathers from all of them at once. Each vertex owns a private weight table at every depth — one weight for each neighbor and each field that neighbor shows it — and shares nothing with any other vertex. The one wrinkle is a lookback window: when a vertex reads from a neighbor, it sees not just that neighbor's newest field but the last few written (a configurable width, two to six). This acts as a short skip connection and keeps the input visible to the early depths.

Training runs the forward pass's loops in reverse, and no depth is spared: the gradient reaches every weight at every depth. For the full story, docs/forward.md and docs/training.md walk through the code loop by loop, with dim-4 examples small enough to check by hand.

Raman baseline extraction (a vibrational spectroscopy application)

The benchmark is the one the sibling projects established: recover the slow fluorescence background under sharp molecular peaks without lifting the baseline into the bands or cutting trenches beneath them. The dataset is 10,000 synthetic LiCoO₂ (lithium cobalt oxide) training spectra and 2,000 held-out validation spectra, scored as RMSE in raw counts.

On this task the frozen-stage siblings — Etalon, WTF, and Cascade, each feeding the same one-layer, one-channel readout — all landed on one floor: 4.76 to 4.82 validation. The LCN, with a dim-11 cube matched to the 2048-bin spectrum and every weight trained, scores 1.98 training / 2.03 validation.

Held-out validation extract, spectra 351 through 354

Grey is the raw spectrum, red the true baseline, blue the extract. At two counts of RMSE the residual is at the scale of the label's own noise, and the red trace all but disappears under the blue. A second experiment widens the cube to dim 12 so half the vertices serve as free hidden units, and scores 1.64 / 1.69. The write-ups are examples/RamanBaseline/ and examples/RamanBaselineNarrowIO/.

Installation

Preferred: install a pre-built wheel from PyPI (no compiler).

pip install hypercube-lcn

import hypercube_lcn as hl
print(hl.__version__)

Package name on PyPI: hypercube-lcn. Import name: hypercube_lcn. Main type: hl.LCN.

Wheels target Python 3.10–3.14 on common Windows, Linux, and macOS machines. Runtime dependency: NumPy only.

From source (full repository)

To compile the extension yourself, clone this entire repository (not a minimal source-only download of the python/ folder alone — the C++ core lives next to python/). You need Python 3.10+, a C++23 compiler, and CMake ≥ 3.20.

git clone https://github.com/dliptak001/HypercubeLCN.git
cd HypercubeLCN/python
pip install .

On Windows with CLion's MinGW, put that compiler's bin folder (and Ninja) on your PATH, then:

pip install . --no-build-isolation --force-reinstall --no-deps

(Exact CLion paths change with the version.)

Quick start

You bring each sample as a length-N float array (N = 2^dim). How you get there — pad an image, reshape a spectrum, invent a layout — is up to you. This package does not pack 784 pixels or 2048 bins for you.

Shapes that matter:

Array	Shape	Notes
`fields`	`(count, N)`	one length-N field per row
`targets`	`(count, width)`	width ≤ N; width < N masks the loss to vertices 0..width-1

import numpy as np
import hypercube_lcn as hl

dim = 6
N = 2**dim
rng = np.random.default_rng(0)
fields = rng.standard_normal((256, N)).astype(np.float32)
targets = 0.5 * (fields + fields[:, np.arange(N) ^ 1])  # a local map

net = hl.LCN(dim=dim, gather_span=3, tanh_last=False,
             lr=1e-2, lr_min_frac=0.05, restore_best=True)
net.fit(fields, targets, epochs=40, batch_size=16, verbose=True)

prediction = net.forward(fields[0])   # (N,) float32
print(net)

net.save("model.pkl")
loaded = hl.LCN.load("model.pkl")

Step by step (same loop, more control)

fit is nothing but this loop — drive it yourself to interleave your own metrics, schedules, or early stopping:

for epoch in range(epochs):
    net.set_epoch(epoch, epochs)          # cosine learning rate
    for start in range(0, count, batch_size):
        net.zero_grad()
        for i in range(start, start + batch_size):
            net.forward(fields[i])
            net.loss(targets[i])          # seeds the gradient
            net.backward()                # accumulates (sums) it
        net.adam()                        # one step per batch
    net.observe(epoch_metric, epoch)      # restore_best bookkeeping
net.restore_best()

The gradient is a sum over the batch, so the effective step scales with batch size — the same convention as the C++ examples.

Features

One class — hypercube_lcn.LCN is the whole product surface
fit — shuffle, batch, cosine schedule, restore-best, in one call
Custom loops — zero_grad / forward / loss / backward / adam / set_epoch exposed one-to-one with the C++ API
dim 4–24 — field length N = 2^dim; depth z_max (default: dim); lookback window gather_span 2–6
Masked loss — targets narrower than N leave the rest of the cube as free hidden units
tanh_last — off by default (raw accumulator out); on confines the output to (-1, 1)
Weights and gradient as NumPy — net.weights (settable) and net.grad, layout depth, axis, tap, vertex
Save / load — save / load (pickle: config + weights; optimizer state is not stored; v1 pickles from 1.1.0 are rejected)
NumPy float32 — arrays converted for you; prefer contiguous float32

Examples

For a first try, paste the Quick start after pip install hypercube-lcn. That is self-contained.

The demo scripts on GitHub under python/examples/ are there to open or download — they are not added to your machine by pip.

Script	What it is for
synthetic_regression.py	Toy field map: `fit` with verbose loss, then held-out MSE

# from a clone of HypercubeLCN, after: pip install hypercube-lcn
python python/examples/synthetic_regression.py

These use easy made-up fields so the API is obvious — not scores to publish.

Documentation

Doc	Role
docs/Python_SDK.md	Canonical Python API — every method, layout, pickle, limits
docs/CPP_SDK.md	Native library guide (same product, C++)
Project README	Product story and C++ demos from the repo root
docs/forward.md	The forward pass, loop by loop, with hand-checkable examples
docs/training.md	Backprop and Adam through the same loops
examples/README.md	The C++ example programs and datasets

Ecosystem

HypercubeCascade: both frozen stages in series + thin readout.
HypercubeCNN: cube-native conv stack with shared kernels.
HypercubeESN: echo-state / reservoir computing on streams.
HypercubeEtalon: frozen etalon transit + thin readout.
HypercubeHopfield: Hopfield-style dynamics on the cube.
HypercubeLCN: the locally connected net, every weight trained.
HypercubeWorldModel: frozen WTF encoder + trained LCN predictor and decoder.
HypercubeWTF: frozen reservoir orbit + thin readout.

License

Apache 2.0.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

hypercube_lcn-1.2.0.tar.gz (26.8 kB view details)

Uploaded Sep 11, 2026 Source

Built Distributions

If you're not sure about the file name format, learn more about wheel file names.

The dropdown lists show the available interpreters, ABIs, and platforms. Enable javascript to be able to filter the list of wheel files.

hypercube_lcn-1.2.0-cp314-cp314-win_amd64.whl (325.5 kB view details)

Uploaded Sep 11, 2026 CPython 3.14Windows x86-64

hypercube_lcn-1.2.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (133.8 kB view details)

Uploaded Sep 11, 2026 CPython 3.14manylinux: glibc 2.27+ x86-64manylinux: glibc 2.28+ x86-64

hypercube_lcn-1.2.0-cp314-cp314-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl (123.5 kB view details)

Uploaded Sep 11, 2026 CPython 3.14manylinux: glibc 2.26+ ARM64manylinux: glibc 2.28+ ARM64

hypercube_lcn-1.2.0-cp314-cp314-macosx_13_0_x86_64.whl (125.0 kB view details)

Uploaded Sep 11, 2026 CPython 3.14macOS 13.0+ x86-64

hypercube_lcn-1.2.0-cp314-cp314-macosx_13_0_arm64.whl (114.8 kB view details)

Uploaded Sep 11, 2026 CPython 3.14macOS 13.0+ ARM64

hypercube_lcn-1.2.0-cp313-cp313-win_amd64.whl (315.6 kB view details)

Uploaded Sep 11, 2026 CPython 3.13Windows x86-64

hypercube_lcn-1.2.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (133.8 kB view details)

Uploaded Sep 11, 2026 CPython 3.13manylinux: glibc 2.27+ x86-64manylinux: glibc 2.28+ x86-64

hypercube_lcn-1.2.0-cp313-cp313-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl (123.0 kB view details)

Uploaded Sep 11, 2026 CPython 3.13manylinux: glibc 2.26+ ARM64manylinux: glibc 2.28+ ARM64

hypercube_lcn-1.2.0-cp313-cp313-macosx_13_0_x86_64.whl (124.8 kB view details)

Uploaded Sep 11, 2026 CPython 3.13macOS 13.0+ x86-64

hypercube_lcn-1.2.0-cp313-cp313-macosx_13_0_arm64.whl (114.6 kB view details)

Uploaded Sep 11, 2026 CPython 3.13macOS 13.0+ ARM64

hypercube_lcn-1.2.0-cp312-cp312-win_amd64.whl (315.5 kB view details)

Uploaded Sep 11, 2026 CPython 3.12Windows x86-64

hypercube_lcn-1.2.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (133.7 kB view details)

Uploaded Sep 11, 2026 CPython 3.12manylinux: glibc 2.27+ x86-64manylinux: glibc 2.28+ x86-64

hypercube_lcn-1.2.0-cp312-cp312-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl (123.0 kB view details)

Uploaded Sep 11, 2026 CPython 3.12manylinux: glibc 2.26+ ARM64manylinux: glibc 2.28+ ARM64

hypercube_lcn-1.2.0-cp312-cp312-macosx_13_0_x86_64.whl (124.7 kB view details)

Uploaded Sep 11, 2026 CPython 3.12macOS 13.0+ x86-64

hypercube_lcn-1.2.0-cp312-cp312-macosx_13_0_arm64.whl (114.5 kB view details)

Uploaded Sep 11, 2026 CPython 3.12macOS 13.0+ ARM64

hypercube_lcn-1.2.0-cp311-cp311-win_amd64.whl (313.4 kB view details)

Uploaded Sep 11, 2026 CPython 3.11Windows x86-64

hypercube_lcn-1.2.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (132.0 kB view details)

Uploaded Sep 11, 2026 CPython 3.11manylinux: glibc 2.27+ x86-64manylinux: glibc 2.28+ x86-64

hypercube_lcn-1.2.0-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl (122.2 kB view details)

Uploaded Sep 11, 2026 CPython 3.11manylinux: glibc 2.26+ ARM64manylinux: glibc 2.28+ ARM64

hypercube_lcn-1.2.0-cp311-cp311-macosx_13_0_x86_64.whl (122.3 kB view details)

Uploaded Sep 11, 2026 CPython 3.11macOS 13.0+ x86-64

hypercube_lcn-1.2.0-cp311-cp311-macosx_13_0_arm64.whl (113.0 kB view details)

Uploaded Sep 11, 2026 CPython 3.11macOS 13.0+ ARM64

hypercube_lcn-1.2.0-cp310-cp310-win_amd64.whl (312.4 kB view details)

Uploaded Sep 11, 2026 CPython 3.10Windows x86-64

hypercube_lcn-1.2.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (131.3 kB view details)

Uploaded Sep 11, 2026 CPython 3.10manylinux: glibc 2.27+ x86-64manylinux: glibc 2.28+ x86-64

hypercube_lcn-1.2.0-cp310-cp310-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl (122.0 kB view details)

Uploaded Sep 11, 2026 CPython 3.10manylinux: glibc 2.26+ ARM64manylinux: glibc 2.28+ ARM64

hypercube_lcn-1.2.0-cp310-cp310-macosx_13_0_x86_64.whl (120.9 kB view details)

Uploaded Sep 11, 2026 CPython 3.10macOS 13.0+ x86-64

hypercube_lcn-1.2.0-cp310-cp310-macosx_13_0_arm64.whl (111.6 kB view details)

Uploaded Sep 11, 2026 CPython 3.10macOS 13.0+ ARM64

File details

Details for the file hypercube_lcn-1.2.0.tar.gz.

File metadata

Download URL: hypercube_lcn-1.2.0.tar.gz
Upload date: Sep 11, 2026
Size: 26.8 kB
Tags: Source
Uploaded using Trusted Publishing? Yes
Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for hypercube_lcn-1.2.0.tar.gz
Algorithm	Hash digest
SHA256	`10f2496547243dcb34b3f4a4fb12ae2f8f44fe26e3b0a5e9a8c2f3be1aa8914e`
MD5	`5fe93ac62b5ac6d6a48a1c0217c6fc84`
BLAKE2b-256	`f444a2efdb0acb87eb936aed451d311c772675861b6e0fdd83e1867a2b81bb6c`

Algorithm	Hash digest
SHA256	`333cdb1cca51939889989a9685a360fd8b997578a32cf68353b443d3792a9ce7`
MD5	`1a306699fceb832325e2ddcc3be7e16e`
BLAKE2b-256	`e9b534c4ce49e6cc78751fece81413497dd0bc01bda45d8003446e6ae3040926`

Algorithm	Hash digest
SHA256	`d2860e6a09f6eafa31d371a3d543b74cadef550a5d24cb3878f65e641d3e1cbc`
MD5	`a3bf24f7bc6935509b458b62169500b3`
BLAKE2b-256	`2737aa3501d20de492e866887b2f17d1124ea5f9835ec79e0a3893dbe1432cfb`

Algorithm	Hash digest
SHA256	`feff1f8b41b498c3dc47418f2382f175904ef94750925d93d8613ca377ed0a76`
MD5	`8305a1edd3bff707026f3776e6ee1d05`
BLAKE2b-256	`1ab2515acb0e068652a4701e88ba3dedb9a720b6e75b267ffe7105d26616f394`

Algorithm	Hash digest
SHA256	`25b8550109bd9d68db5a4d352fe02faeed59c1e69c59ceef771719d0eb813d3c`
MD5	`d955cd381cb84cb4f23cf2fdaf0bc3f9`
BLAKE2b-256	`c41a41ec2a527abfdae4ea6e5ec5023f4ddc4a3596cfda38ba3e0e90c1d41add`

Algorithm	Hash digest
SHA256	`2bc8837d0e6c21fb3dafa230dc3911922e4279c4314da2ee0f99eb1161cb7db6`
MD5	`310e5dd7702bbc7c1c14f766b8dcb1ba`
BLAKE2b-256	`4df6bc86c61bbd39c3222ae2ba769715671cd344aede2d557217fc6dddcb6b01`

Algorithm	Hash digest
SHA256	`9c8d4ad14e6db85057c36c95d7c2938cfaaea502d0c8a284e676533fb8581a31`
MD5	`f2cdc33a3f61fb8150923e49ea2a45eb`
BLAKE2b-256	`1d6bf349d1c042fc64c2b1b35a1cd17990e09252f7024b1d036e1c87489b14a3`

Algorithm	Hash digest
SHA256	`d97513517398e6648d5eb467736a8cef040cf7d125794195ff2c9acb882e758d`
MD5	`5553f956a612b7f26fcba48120f92c73`
BLAKE2b-256	`13d8af32ae1532759300218c090c3b6c0f2a94f480f36d4aeabd6f63b4217985`

Algorithm	Hash digest
SHA256	`2cfd2d04abef741b3c724620001e7f189cc6f044a5d4633e17eae7e9beee0250`
MD5	`0442100e79e7c4a2719177d5efa07bee`
BLAKE2b-256	`f7886d023104bfb739331225de3b91fa4b6f02c24ebfbc67eb4f71318e677ace`

Algorithm	Hash digest
SHA256	`b59776b48fe61740e7552f76d093eec9afcddf3d254c8f5bc4bf9787a009b7b8`
MD5	`f5f6f25980d0d7c70ec355428d0691b6`
BLAKE2b-256	`75cd7f7a9c487e58cbdd4a6d432f18d14c91d3ac7100ca6a5edb64d85789d0cf`

Algorithm	Hash digest
SHA256	`c584a10dd93735c3a1576f45daa55ef2d178ea3bb9bcdf7aa82acdcd098ff3b1`
MD5	`fd6cb10d6d01ed2c51978105b933c9dd`
BLAKE2b-256	`46e5fbfcf5cf8dff6c2267ff8dc3d630c323d86f78eb5cb96aef62fa3c82b020`

Algorithm	Hash digest
SHA256	`7db8cf1c30bdb74668c334a7f11f145755fdd418579fd67d96b81a61009f1ccc`
MD5	`6dcc92b4297e9ba4ab3a1f7d116cb478`
BLAKE2b-256	`a1757d1a2a23ed105e148e178c6f7360f7714b2d39823c8d97c8b6cbe6c54b72`

Algorithm	Hash digest
SHA256	`db2ef118ea877f5c0caaa45dc00d79ba9c19ce566da5f03726ea94073283529d`
MD5	`d2f58b1b6ab17b6fa402f15fa3001e19`
BLAKE2b-256	`67160dca219428880bce4fbb027da2e9f69bb697a3658d6ff058c269a2e6c4e0`

Algorithm	Hash digest
SHA256	`54aca799b61e0af949c8791f63ef94ba3f7be539db4889f8a5d9cb2763d27a9d`
MD5	`d9418224f0dcb280c001422efc4d59eb`
BLAKE2b-256	`aa34a2492fca14b59437abdf9286693494b7ed15b9a3f3d7e8ccd535cf929dd8`

Algorithm	Hash digest
SHA256	`e0bc924e8e68c47b4f876b4113eaceffc57563edec0a5c213cb3d0f2cc9a39d8`
MD5	`c711e4e556974ca0f0a78837fc8dddcd`
BLAKE2b-256	`f17afb44a795addf732fa095cf6266bc6d8d7cc7dec75ee3a2089c1f4baaf3f4`

Algorithm	Hash digest
SHA256	`a1e3c9924fd78fc03f6a024293bd388d6023e58632137ae80b7bd97e277dc707`
MD5	`3c38586b7c5a2ff689e2f71329575b12`
BLAKE2b-256	`6300a86e60c10eb1cb0dd867388df7e1871d7f69251553f7e63dc93aea1701df`

Algorithm	Hash digest
SHA256	`1ebfc43dd9b84d7d2ab76b3c22524fdb5e9102100240d5961ed114afd2565fee`
MD5	`38692cc36d41195a10b9f4c2f118599e`
BLAKE2b-256	`2b729648496e0a5aa377029a52edf32c98211de41f34b36a5f1d9dd51e65989a`

Algorithm	Hash digest
SHA256	`bb8627091ce39bc1967998b0ca25b7430f7e0296680a43972be5d81c92fef5b5`
MD5	`e74c57dfded31db7aa5606e4f13d5d59`
BLAKE2b-256	`93fab249820cd85d7812f914c65df5aaf998b9a39376cd2ca290087ea74a71cb`

Algorithm	Hash digest
SHA256	`6c530c21fa28657ece4323ed8034e67cdd8abd6af33e2f8a12d5af6ba2223219`
MD5	`1a73780fef1b98bfe3d0c18fc8bc8d37`
BLAKE2b-256	`2fff29932be0a15bdcdbec7dbe4c17ce3f04aec53335b5e2a26eeead1522a7de`

Algorithm	Hash digest
SHA256	`dff5fc2f810fd9ce98430a5479249b265434aa13f3c21f49594427bbadec0215`
MD5	`89bb313a62633b1a7c9666cfc43657f4`
BLAKE2b-256	`e23e39261dc6d196a72237962449e198419d7466f02f6ec51770de60a17154ad`

Algorithm	Hash digest
SHA256	`2fd381212dcfd1cefc1ab9c0b72235f269110d3c793a8eb0ff55e45c6ed92440`
MD5	`90798a1c7de37cd5dc0375110c21b9d0`
BLAKE2b-256	`5ce8dcfa01eea83cb47ece21cdd275339a625e2978ebf89dbc9a13749a5d6e23`

hypercube-lcn 1.2.0

Hypercube LCN

Definitions

The LCN

Raman baseline extraction (a vibrational spectroscopy application)

Installation

From source (full repository)

Quick start

Step by step (same loop, more control)

Features

Examples

Documentation

Ecosystem

License

Project links

Key dates

1 maintainer

GitHub Statistics

License expression

Requires

Provides Extra

Classifiers

Download files

Source Distribution

Built Distributions

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Algorithm	Hash digest
SHA256	`0a2eab3f3cab6b235ac3626dd64441b36bc92c5c5e1384022e5fdbfec5672966`
MD5	`5fa04bded56cff13f48ceac5456b789c`
BLAKE2b-256	`8b61fbe30e8e4559adb6cb1b04765d513592ae6302605165fc5e519ac360ef53`

Algorithm	Hash digest
SHA256	`0b7fe7db8c770b6a2d679c7f1767d5f03d68c7135ea9b4cbfa5faa37b3d30206`
MD5	`b61d3857e58cfd3d4f108a6bbe35a263`
BLAKE2b-256	`c2116d1a065a226613adbfc3acf02f4a1f605ec3321e1135a5d936b86c0ff31c`

Algorithm	Hash digest
SHA256	`e0b34b0fd72ee65221cc42ceec260c85998166db1ff9108b585feab02afc2df4`
MD5	`3f7b0bfd0cbd6b46b3d65be24a6b3c31`
BLAKE2b-256	`a7a5b035d96ae7ac4ebe0b74fcc6089df4c44230b364df13b261b529d53b61f3`

Algorithm	Hash digest
SHA256	`9731a4bc67dd6d327919f9f74f57d73e090a42f233ee100e2142e6adb67fd8e2`
MD5	`048855aa59c609e6aa2764f4a69334af`
BLAKE2b-256	`dfae4349b8f0c730efe808297c2e7d67c97bf2372b725da96d17a08074bbc6b7`

Algorithm	Hash digest
SHA256	`5b9666cf615a4d0bdd31aeaf0ff931292e0d43050f5c9b2b16b4f80c39cbf5ef`
MD5	`1f64b3aaf96d417adbcb10aa478412b4`
BLAKE2b-256	`c1763395c27385fb3769cc4990ac56059d617aeb49464106f57004bb08a99ec1`