Skip to main content

rumi (python)

Python bindings for Rumi, stateless raster storage for AI4EO. Rumi stores images as (B, Y, X) and time series as (T, B, Y, X).

pip install rumi-eo

Rumi currently supports Linux and macOS and requires Python 3.11 or newer. The format and APIs may change before 1.0. See the project README for examples and the format specification for the wire format.

Reads return NumPy by default. PyTorch, JAX and TensorFlow results use DLPack; framework="dlpack" returns the one-shot producer itself. Compatibility is checked before opening the source, and Rumi never casts a dtype to make it fit.

PyTorch can read straight into a CUDA tensor on Linux or an MPS tensor on macOS:

chip = rumi.read(path, header, framework="torch", device="cuda:0",
                 window=(row, column, height, width))

mps_chip = rumi.read(path, header, framework="torch", device="mps",
                     window=(row, column, height, width))

cpu_decoded = rumi.read(path, header, framework="torch", device="mps",
                         decode_backend="cpu",
                         window=(row, column, height, width))

The header, byte ranges and copy plan stay on the CPU. PFOR and lanes frames of one- and two-byte samples decode on the device when they have one chunk; other frames use CPU staging before GeoZL writes the result to the selected GPU. decode_backend="cpu" forces complete CPU decoding for all selected frames; "auto" is the default. MPS reads require PyTorch 2.10 or newer. GPU frame planning follows rumi.set_num_threads or RUMI_NUM_THREADS, set before the first read, and defaults to one thread per process. R and Julia device arrays are not implemented yet.

Metadata

Release files for rumi-eo 0.30.0

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

Built distributions (wheels)

Table of built distributions (wheels) for rumi-eo 0.30.0
File Interpreter ABI Platform Actions
rumi_eo-0.30.0-py3-none-manylinux_2_28_x86_64.whl Python 3 none Linux glibc 2.28+ x86-64 Details for rumi_eo-0.30.0-py3-none-manylinux_2_28_x86_64.whl
rumi_eo-0.30.0-py3-none-macosx_11_0_arm64.whl Python 3 none macOS 11.0+ ARM64 Details for rumi_eo-0.30.0-py3-none-macosx_11_0_arm64.whl

Total release size: 17.7 MB

Release files / rumi_eo-0.30.0-py3-none-manylinux_2_28_x86_64.whl

Download URL rumi_eo-0.30.0-py3-none-manylinux_2_28_x86_64.whl
Size 13.0 MB
Tags Linux glibc 2.28+ x86-64 Python 3
SHA-256 checksum
How to use checksums
2262dde582c82a00ef3c5ea87d28dda67ae38cbb27a684e3c1ebf5df14ef1a44
BLAKE2b-256 checksum
How to use checksums
3aaa3083e46e24309e9acc5943ef6490004decc14c1c6dccae8ad25b17ffc59f
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 Nov 11, 2026.

Transparency log

Release files / rumi_eo-0.30.0-py3-none-macosx_11_0_arm64.whl

Download URL rumi_eo-0.30.0-py3-none-macosx_11_0_arm64.whl
Size 4.8 MB
Tags Python 3 macOS 11.0+ ARM64
SHA-256 checksum
How to use checksums
cb9bc26193e1566972b338273bdf5350c10b23aeb9f4b5094224ee9f5573d285
BLAKE2b-256 checksum
How to use checksums
6f7bf138106d99073a2eeb8870b49d7e68d8f2595276322f4559d46eb0156f6b
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 Nov 11, 2026.

Transparency log

Security for rumi-eo 0.30.0

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 rumi_eo-0.30.0-py3-none-manylinux_2_28_x86_64.whl

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 Nov 11, 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.

Release history Release notifications | RSS feed

This release

0.30.0 This release

2 release files Full provenance

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