-
-
Notifications
You must be signed in to change notification settings - Fork 59
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Propose new environmental consideration information for ML models #395
Conversation
Note: some sources use "CO2e" and some "CO2eq"... gov. agencies tend to use the former, where model providers often use the latter. For example:
Articles often state that they "can be used interchangeably": |
Another reference from tc-54: https://www.whitehouse.gov/wp-content/uploads/2021/10/us-long-term-strategy.pdf |
@jkowalleck @stevespringett |
My presentation from the 03-07-2024 Ecma TC-54 call: |
@mrutkows is this PR related to an existing issue? |
Could we make this a bit more generic so that it could be attached to services, manufacturers, etc? Initially, it could be a type in external reference since the analysis is usually very specific to the needs of a particular organization. |
@mrutkows I think we have consensus to use the current model card approach to adding support for environmental concerns and then in v1.7, we can expand that support to every component and service. What is the status of this PR? We likely have one week to flush this out, otherwise it will need to be postponed to v1.7. |
I still have questions and concerns. I vote for postponing and continued discussions in #396 |
re: #395 (comment) |
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
Signed-off-by: Jan Kowalleck <jan.kowalleck@gmail.com>
@mrutkows here is the neeed fix for the examples: mrutkows#3 as soon as this is merged, we should be golden. |
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
Energy fix examples
Thanks Jan!!! BTW, there is still something wrong in the XSD in that I should have an |
…type in JSON Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
looks good overall.
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
Signed-off-by: Matt Rutkowski <mrutkows@us.ibm.com>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This looks good @mrutkows . Thanks for all the hard work on this. If you can sort out the other
discrepancy, we can get this approved.
## Added * Core enhancement: Attestation ([#192](#192) via [#348](#348)) * Core enhancement: Cryptography Bill of Materials — CBOM ([#171](#171), [#291](#291) via [#347](#347)) * Feature to express the URL to source distribution ([#98](#98) via [#269](#269)) * Feature to express the URL to RFC 9116 compliant documents ([#380](#380) via [#381](#381)) * Feature to express tags/keywords for services and components (via [#383](#383)) * Feature to express details for component authors ([#335](#335) via [#379](#379)) * Feature to express details for component and BOM manufacturer ([#346](#346) via [#379](#379)) * Feature to express communicate concluded values from observed evidences ([#411](#411) via [#412](#412)) * Features to express license acknowledgement ([#407](#407) via [#408](#408)) * Feature to express environmental consideration information for model cards ([#396](#396) via [#395](#395)) * Feature to express the address of organizational entities (via [#395](#395)) * Feature to express additional component identifiers: Universal Bill Of Receipts Identifier and Software Heritage persistent IDs ([#413](#413) via [#414](#414)) ## Fixed * Allow multiple evidence identities by XML/JSON schema ([#272](#272) via [#359](#359)) This was already correct via ProtoBuff schema. * Prevent empty `license` entities by XML schema ([#288](#288) via [#292](#292)) This was already correct in JSON/ProtoBuff schema. * Prevent empty or malformed `property` entities by JSON schema ([#371](#371) via [#375](#375)) This was already correct in XML/ProtoBuff schema. * Allow multiple `licenses` in `Metadata` by ProtoBuff schema ([#264](#264) via [#401](#401)) This was already correct in XML/JSON schema. ## Changed * Allow arbitrary `$schema` values by JSON schema ([#402](#402) via [#403](#403)) * Increased max length of `versionRange` (via [`3e01ce6`](3e01ce6)) * Harmonized length of `version` (via [#417](#417)) ## Deprecated * Data model "Component"'s field `author` was deprecated. (via [#379](#379)) Use field `authors` or field `manufacturer` instead. * Data model "Metadata"'s field `manufacture` was deprecated. ([#346](#346) via [#379](#379)) Use "Metadata"'s field `component`'s field `manufacturer` instead. - for XML: `/bom/metadata/component/manufacturer` - for JSON: `$.metadata.component.manufacturer` - for ProtoBuf: `Bom:metadata.component.manufacturer` ## Documentation * Centralize version and version-range (via [#322](#322)) * Streamlined SPDX expression related descriptions (via [#327](#327)) * Enhanced descriptions of `bom-ref`/`refType` ([#336](#336) via [#344](#344)) * Enhanced readability of enum documentation in JSON schema ([#361](#361) via [#362](#362)) * Fixed typo "compliment" -> "complement" (via [#369](#369)) * Added documentation for enum "ComponentScope"'s values in JSON schema ([#293](#293) via [`d92e58e`](d92e58e)) Texts were a taken from the existing ones in XML/ProtoBuff schema. * Added documentation for enum "TaskType"'s values ([#245](#245) via [#377](#377)) * Improve documentation for data model "Metadata"'s field `licenses` ([#273](#273) via [#378](#378)) * Added documentation for enum "MachineLearningApproachType"'s values ([#351](#351) via [#416](#416)) * Rephrased some texts here and there. ## Test data * Added test data for newly added use cases * Added quality assurance for our ProtoBuf schemas ([#384](#384) via [#385](#385))
The fact that datasets used to train AI models are increasingly large and take an enormous amount of energy (and indirectly produce large CO2 emissions) to develop, train and run has come to the forefront. This PR contains proposed additions to the "modelCard" type to account for these considerations when selecting/utilizing a model.
ModelCardConsiderations.environmentalConsiderations
this fixes Propose new environmental consideration information for ML models #396
OrganizationalEntity.address
TODO