Skip to main content

nudemos

The Nu demos as nuspace Planes + creates, under Demos:

  • Counter: a count ticking once a second, kept in its cell's state.
  • Sampled: a series that grows forever, charted from a reservoir sample.
  • Movies: a film base where every film and person is its own live page.
  • MNIST: train a small CNN, steer it while it trains, and track every run. Needs the mnist extra.
  • Memes: a live terminal over the meme market on Solana: launches, trades and every coin's page, synced from public mainnet. Needs the memes extra.
  • Agents: Claude Code sessions run, watched, steered and orchestrated from a space. One agent is one Claude Code session in a page, driven by a short Nu loop; fan-out with a judge, chains and task boards are a few lines of Nu each. Needs the agents extra and a logged-in claude CLI, no API key.
  • Nu demos: all three at once, nested under one plane.
pip install nudemos

nuspace finds it on install through the nuspace.extensions entry point. You don't import it yourself.

MNIST

Launch picks the knobs (optimizer, learning rate, batch size, channels, epochs, dropout) and starts a run: a page, and a headless job under it that trains in its own process. A run page shows the curves, twelve test digits and what the model reads in each, and every layer's weights and gradients over training. Drag the learning rate while it trains, stop it, resume it. Compare lines the runs up. Read puts a run's model to work: pick a run and its latest or best checkpoint, draw a digit, and it reads it on every stroke, showing the 28x28 image the model sees. Pick a run that is still training, or just launch one (the page follows the newest run until you pick), and the same drawing is read again as each checkpoint lands.

Nu owns the loop, torch owns the step: the job cell is a Nu loop over a trainer resource, reading its learning rate from state each chunk. Close the space mid run and open it again: the run carries on from its last checkpoint.

Models live in the space, not in files: each run keeps its latest checkpoint (what it resumes from) and the weights at its best test score, in its own state.

pip install "nudemos[mnist]"

MNIST downloads once, on the first run, into ~/.cache/nuverse-mnist ($NUVERSE_MNIST_DATA to move it). Runs train on cuda, or mps on Apple silicon, else the cpu ($NUVERSE_MNIST_DEVICE to pick). Three epochs take about a minute on a laptop cpu.

Metadata

Release files for nudemos 0.1.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for nudemos 0.1.0
File Size Uploaded Provenance Actions
nudemos-0.1.0.tar.gz 110.7 kB With provenance Details for nudemos-0.1.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for nudemos 0.1.0
File Interpreter ABI Platform Actions
nudemos-0.1.0-py3-none-any.whl Python 3 none any Details for nudemos-0.1.0-py3-none-any.whl

Total release size: 246.6 kB

Release files / nudemos-0.1.0.tar.gz

Download URL nudemos-0.1.0.tar.gz
Size 110.7 kB
Tags Source
SHA-256 checksum
How to use checksums
e0d78b5508824cc9e041ef32ea611f339f5460fda32c932de44802a765e53b98
BLAKE2b-256 checksum
How to use checksums
beba773b7733a6ca137565ef4dfb1b9f2179cac6a8b754f972b12358d18c3024
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 16, 2026.

Transparency log

Release files / nudemos-0.1.0-py3-none-any.whl

Download URL nudemos-0.1.0-py3-none-any.whl
Size 135.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
acf5c184903922c9e5832c47bbf26a5382b00d01f937dd124b9caa41c8b3fae7
BLAKE2b-256 checksum
How to use checksums
fb7e24aff112865d7934e8a15a51b0ba9386c85f2b999f27462cd00af74aa58c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 16, 2026.

Transparency log

Security for nudemos 0.1.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

Both files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for nudemos-0.1.0.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 16, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page