Skip to main content

astro-weights logo: astronaut with a gold visor and a violet chest patch

astro-weights

Small course corrections for open LLMs.
Agent-driven weight-space edits, exported as standard LoRA adapters.

PyPI version Python versions CI

Docs · PyPI · Changelog · Contributing

The goal: an agent proposes small, targeted low-rank edits to a model's weights, and the ones that hold up under paired tests are saved as plain PEFT LoRA adapters. Today (0.0.8) the library covers the edit core (build low-rank edits, save and load them as LoRA adapters, apply them to transformers models), difference-of-means directions with a split-half gate, and paired eval statistics. LoRA is the output format because every serving stack already loads it (transformers, PEFT, vLLM, llama.cpp and the rest), so an edit needs no custom runtime: ship the adapter, load it like any other, and drop it to get the base model back.

Status: pre-alpha. 0.0.5 is on PyPI. The API will change without notice until 1.0.

Install

pip install astro-weights

To apply edits to transformers models, add the hf extra (pulls in torch, transformers and peft):

pip install "astro-weights[hf]"

Quick look

Build a rank-1 edit with numpy and save it as a LoRA adapter:

import numpy as np
from astro_weights.edits import Edit, masked_rank1
from astro_weights.export import save_peft

W0 = np.random.default_rng(0).standard_normal((16, 48), dtype=np.float32)
u = np.eye(16, dtype=np.float32)[0]
mask = np.ones(48, dtype=np.float32)
delta = masked_rank1("model.layers.1.mlp.down_proj", W0, u, mask, alpha=0.5)
save_peft(Edit(deltas=(delta,), base_model="org/name"), "my-edit")

Summarize and verify the saved adapter:

astro-weights inspect my-edit

Apply it to a transformers model (hf extra), and remove it again:

from astro_weights.export import load_peft
from astro_weights.runners.hf import apply, remove

peft_model = apply(model, load_peft("my-edit"))  # model: a transformers causal LM
model = remove(peft_model)  # base weights were never modified

apply also finds layers named differently, as in text-only and multimodal loads of one checkpoint: see How-to: apply across model classes.

Development

git clone https://github.com/astro-weights/astro-weights
cd astro-weights
uv sync
uv run pytest

See CONTRIBUTING.md for the full set of checks and how to send a pull request.

Citation

If you use astro-weights in your work, please cite it. The metadata is in CITATION.cff; on GitHub, use Cite this repository in the sidebar.

License

Apache-2.0.

The logo is based on the Noto Emoji "person astronaut" glyph, Copyright Google LLC, Apache-2.0. See docs/assets/LOGO-LICENSE.md.

Metadata

Release files for astro-weights 0.0.8

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

Source distribution (sdist)

Source distribution for astro-weights 0.0.8
File Size Uploaded Provenance Actions
astro_weights-0.0.8.tar.gz 30.9 kB With provenance Details for astro_weights-0.0.8.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for astro-weights 0.0.8
File Interpreter ABI Platform Actions
astro_weights-0.0.8-py3-none-any.whl Python 3 none any Details for astro_weights-0.0.8-py3-none-any.whl

Total release size: 67.9 kB

Release files / astro_weights-0.0.8.tar.gz

Download URL astro_weights-0.0.8.tar.gz
Size 30.9 kB
Tags Source
SHA-256 checksum
How to use checksums
21661ce94b9192f878b2726d09de79745ea8d435f9ccf5ae6fc05a1139e7ba9e
BLAKE2b-256 checksum
How to use checksums
bbc341c90fbe6ce07a640f016678c05b4ab88abb29ebe440a431a1b21a496790
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

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 15, 2026.

Transparency log

Release files / astro_weights-0.0.8-py3-none-any.whl

Download URL astro_weights-0.0.8-py3-none-any.whl
Size 37.0 kB
Tags Python 3
SHA-256 checksum
How to use checksums
73a3a64592c745c0cf54f4f7a1cc56f6cc08a4da6b7004f48aaff30e165429bc
BLAKE2b-256 checksum
How to use checksums
958a4f616ac86ce9b505c9e233427a00667e286e5b3e6cb6e69403378d8d3b9a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

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 15, 2026.

Transparency log

Security for astro-weights 0.0.8

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 astro_weights-0.0.8.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 Nov 15, 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