Dialz: A Python Toolkit for Steering Vectors
Docs: https://cardiffnlp.github.io/dialz/
Paper: https://www.arxiv.org/abs/2505.06262
About
Steering vectors allow users to modify activations at inference time to amplify or weaken a 'concept', e.g. honesty or positivity.
Dialz supports a diverse set of tasks, including creating contrastive pair datasets, computing and applying steering vectors, and visualizations.
A basic tutorial can be found here.
Installation
pip install dialz
Check out the full documentation for usage information.
Contributing
Any contributions to improve this project are welcome! Please open an issue or pull request in this repo with any changes you have.
License
This code is released under a MIT license.
1 maintainer
Metadata
1 maintainer
Release files for dialz 1.1.4
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| dialz-1.1.4.tar.gz | 3.9 MB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| dialz-1.1.4-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 7.8 MB
Release files / dialz-1.1.4.tar.gz
| Download URL | dialz-1.1.4.tar.gz |
|---|---|
| Size | 3.9 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
978b4f4eda902eee2f194bb2bbad7b4357f1b9b18c478427c1c87fbd3533983f
|
|
BLAKE2b-256 checksum How to use checksums |
7d02e51d80d8ce867dc82bd82b710163fd235397d9b6014eceee4f9364a3f7e8
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.12.10
|
Release files / dialz-1.1.4-py3-none-any.whl
| Download URL | dialz-1.1.4-py3-none-any.whl |
|---|---|
| Size | 3.9 MB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
447ec84cbb25ea2f63ff583db1d75af3e84b922090e7f14434202fdcf53519c7
|
|
BLAKE2b-256 checksum How to use checksums |
e33774a2893dc6218d2ad0768a1b9f6eb15a34d20f3badb307ef37c26c511e32
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.12.10
|
