forked from ossf/tac
-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Create sandbox project for the model signing work
We have a working group that meets for model signing work, as part of ossf/ai-ml-security#10. Since this working group helps in developing https://github.com/sigstore/model-transparency and building standards around it, it needs to be officially a project, not a WG. We add the project at a sandbox stage. Please note that there are 2 repos invovled here: - https://github.com/sigstore/model-transparency which will be owned by Sigstore and is just the implementation work for the library for model signing - a new repository to be created under https://github.com/ossf to represent standard documents, as outputs of this project These two repositories should work in unison to achieve a common goal. Signed-off-by: Mihai Maruseac <mihaimaruseac@google.com>
- Loading branch information
1 parent
1556885
commit 803e00d
Showing
2 changed files
with
98 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
97 changes: 97 additions & 0 deletions
97
process/project-lifecycle-documents/model_signing_sandbox_stage.md
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,97 @@ | ||
## Application for creating a new project at Sandbox stage | ||
|
||
### List of project maintainers | ||
|
||
The project has 4 maintainers from 3 different organizations: | ||
|
||
* Laurent Simon, Google, @laurentsimon | ||
* Daniel Major, NVidia, | ||
* Eoin Wickens, HiddenLayer, @EWickens | ||
* Mihai Maruseac, Google, @mihaimaruseac | ||
|
||
### Mission of the project | ||
|
||
The project must be aligned with the OpenSSF mission and either be a novel | ||
approach for existing areas, address an unfulfilled need, or be initial code | ||
needed for OpenSSF WG work. It is preferred that extensions of existing OpenSSF | ||
projects collaborate with the existing project rather than seek a new project. | ||
|
||
> Create a cryptographic signing specification for artificial intelligence and | ||
> machine learning models, addressing challenges such as very large models that | ||
> can be used separately, and the signing of multiple disparate file formats | ||
> held within a directory. This specification may have wider applicability to | ||
> signing directories of multiple arbitrary file formats. This specification may | ||
> later be proposed as a formal standard. | ||
#### Specific Goals Include: | ||
|
||
* Develop standards for efficient hashing of large models | ||
* Develop standards for efficient verification of models that contain multiple | ||
formats in the same place | ||
* Develop standards for efficient verification of models at inference time. | ||
|
||
To achieve these goals, we work on developing | ||
https://github.com/sigstore/model-transparency library as an OSS standard for ML | ||
models signing and verification. The aim of this project is to guide the | ||
`model-transparency` development and help in standardizing hashing, | ||
verification, and deployment (e.g., model signature format). | ||
|
||
#### Non-Goals Include: | ||
|
||
* Developing a new model format | ||
* Handling security of ML outputs | ||
|
||
### OpenSSF Mission Alignment | ||
|
||
We believe our mission aligns with the OpenSSF mission in the following ways: | ||
|
||
> make it easier to sustainably secure the development, maintenance, and | ||
> consumption of the open source software | ||
The model signing project aims to reduce the complexity of signing and verifying | ||
the integrity of models, making it easier to be adopted by the industry at | ||
large. | ||
|
||
> fostering collaboration | ||
The library is developed by a cohort of independent companies working together | ||
to solve common problems. The goal is to integrate the library with most tools | ||
that ML practitioners use, to uplift the entire ecosystem. | ||
|
||
> establishing best practices | ||
The model signing library must be strongly tested. It should define standards | ||
for efficiently hashing and verifying integrity of models. | ||
|
||
### AI/ML Security WG Alignment | ||
|
||
This project started in parallel with the AI/ML Security WG. During one meeting | ||
of the WG, it was decided to spin up a SIG for model signing | ||
(https://github.com/ossf/ai-ml-security/issues/10). Since the output of this SIG | ||
is in code for this library and associated standards and specs, we need to make | ||
this a project. | ||
|
||
### IP policy and licensing due diligence | ||
|
||
When contributing an existing Project to the OpenSSF, the contribution must | ||
undergo license and IP due diligence by the Linux Foundation (LF). | ||
|
||
* Yes: | ||
* Library code under | ||
[`model_transparency`](https://github.com/sigstore/model-transparency) is | ||
part of Sigstore, which is already an OpenSSF (thus, LF) member | ||
* Standardization work, etc. will occur under a new repository to be created | ||
under OpenSSF | ||
|
||
### Project References | ||
|
||
The project should provide a list of existing resources with links to the repository, and if available, website, a roadmap, contributing guide, demos and walkthroughs, and any other material to showcase the existing breadth, maturity, and direction of the project. | ||
|
||
| Reference | URL | | ||
|--------------------|-----| | ||
| Main Repository | TODO | | ||
| Contributing guide | TODO | | ||
| Security.md | TODO | | ||
| Roadmap | TODO | | ||
| Demos | TODO | | ||
| Other | TODO | |