(New construct): Add support to deploy custom models to a sagemaker endpoint #150
Assignees
Labels
Comments
krokoko
added
documentation
krokoko
added
effort/medium
krokoko
added a commit
that referenced
this issue
Jan 12, 2024
…emaker (#180) * feat(construct): New construct CustomSagemakerEndpoint Fixes (New construct): Add support to deploy custom models to a sagemaker endpoint #150 * chore(construct): Update documentation * feat(construct): Add new projen task to run the code generation step (create list of available containers / jumpstart models). Can be used through projen generate-models-containers * chore(construct): Update list of available models * fix(construct): Fixes (gen-ai): JumpStartSageMakerEndpoint not working #179 : address the issue of some models from JumpStart which can't be deployed. This is due to the construct requiring model data to be provided as a compressed object, however some newer JumpStart models provide uncompressed model artifacts. Tested the fix with the reported model - whisper base 2.0.0. Deployed model and ran inference. This requires an upgrade of cdk as the update to the L1 CloudFormation resource was added recently (addition of the property https://docs.aws.amazon.com/cdk/api/v2/docs/aws-cdk-lib.aws_sagemaker.CfnModel.ModelDataSourceProperty.html)
oconpa
pushed a commit
to oconpa/generative-ai-cdk-constructs
that referenced
this issue
Jan 16, 2024
…emaker (awslabs#180) * feat(construct): New construct CustomSagemakerEndpoint Fixes (New construct): Add support to deploy custom models to a sagemaker endpoint awslabs#150 * chore(construct): Update documentation * feat(construct): Add new projen task to run the code generation step (create list of available containers / jumpstart models). Can be used through projen generate-models-containers * chore(construct): Update list of available models * fix(construct): Fixes (gen-ai): JumpStartSageMakerEndpoint not working awslabs#179 : address the issue of some models from JumpStart which can't be deployed. This is due to the construct requiring model data to be provided as a compressed object, however some newer JumpStart models provide uncompressed model artifacts. Tested the fix with the reported model - whisper base 2.0.0. Deployed model and ran inference. This requires an upgrade of cdk as the update to the L1 CloudFormation resource was added recently (addition of the property https://docs.aws.amazon.com/cdk/api/v2/docs/aws-cdk-lib.aws_sagemaker.CfnModel.ModelDataSourceProperty.html)
oconpa
pushed a commit
to oconpa/generative-ai-cdk-constructs
that referenced
this issue
Jan 16, 2024
feat: combining all parent 364274d author github-actions <github-actions@github.com> 1702080482 +0000 committer Patrick <ocopatr@amazon.com> 1705428058 -0800 chore(deps): upgrade dependencies Upgrades project dependencies. See details in [workflow run]. [Workflow Run]: https://github.com/awslabs/generative-ai-cdk-constructs/actions/runs/7147565079 ------ *Automatically created by projen via the "upgrade-main" workflow* Signed-off-by: github-actions <github-actions@github.com> chore(deps): upgrade dependencies Upgrades project dependencies. See details in [workflow run]. [Workflow Run]: https://github.com/awslabs/generative-ai-cdk-constructs/actions/runs/7161144540 ------ *Automatically created by projen via the "upgrade-main" workflow* Signed-off-by: github-actions <github-actions@github.com> chore(doc): update documentation and generate new list of models to deploy chore(deps): upgrade dependencies Upgrades project dependencies. See details in [workflow run]. [Workflow Run]: https://github.com/awslabs/generative-ai-cdk-constructs/actions/runs/7174960686 ------ *Automatically created by projen via the "upgrade-main" workflow* Signed-off-by: github-actions <github-actions@github.com> feat: qa concurrency props fix(constructs): fix log group name fix(lint): run linter fix(lint): differentiate name feat(doc): update readme with additional resources chore(deps): upgrade dependencies Upgrades project dependencies. See details in [workflow run]. [Workflow Run]: https://github.com/awslabs/generative-ai-cdk-constructs/actions/runs/7228300421 ------ *Automatically created by projen via the "upgrade-main" workflow* Signed-off-by: github-actions <github-actions@github.com> chore(deps): upgrade dependencies Upgrades project dependencies. See details in [workflow run]. [Workflow Run]: https://github.com/awslabs/generative-ai-cdk-constructs/actions/runs/7255404764 ------ *Automatically created by projen via the "upgrade-main" workflow* Signed-off-by: github-actions <github-actions@github.com> chore(deps): upgrade dependencies Upgrades project dependencies. See details in [workflow run]. [Workflow Run]: https://github.com/awslabs/generative-ai-cdk-constructs/actions/runs/7325128447 ------ *Automatically created by projen via the "upgrade-main" workflow* Signed-off-by: github-actions <github-actions@github.com> chore: upgrading package versions Docs no longer support langchain 0.0.329 Signed-off-by: Patrick O'Connor <35761519+oconpa@users.noreply.github.com> fix(config): update .projenrc.ts Add IDE specific ignore folders Signed-off-by: Michael Walker <michaelhuytran@gmail.com> feat(configs): add gitignore for IDE files ci(projen): uncapping cdk version to allow usage with latest cdk less pip error ci(projen): lowest build version feat(layer): additional packages prop, rename of layers for ease, typing from layer construct fix: remove console log statements Signed-off-by: Scott Schreckengaust <scottschreckengaust@users.noreply.github.com> feat(constructs): add how to section for testing constructs feat(docs): add screenshots fix(readme): update README.md Co-authored-by: Scott Schreckengaust <scottschreckengaust@users.noreply.github.com> Signed-off-by: Michael Walker <michaelhuytran@gmail.com> fix(readme): update README.md Co-authored-by: Scott Schreckengaust <scottschreckengaust@users.noreply.github.com> Signed-off-by: Michael Walker <michaelhuytran@gmail.com> fix(readme): update README.md Co-authored-by: Scott Schreckengaust <scottschreckengaust@users.noreply.github.com> Signed-off-by: Michael Walker <michaelhuytran@gmail.com> feat(docs): move to developer guide feat(docs): move to developer guide fix: ts-jest configuration under globals Projen 0.78.9 fixes the warning: > (WARN) Define `ts-jest` config under `globals` is deprecated Signed-off-by: Scott Schreckengaust <scottschreckengaust@users.noreply.github.com> fix: correct opt-out logic (awslabs#183) fix: correct opt-out logic * dynamically fetch version from the package.json * include the name of the class * string format * camel case solution_id variable and move into context --------- Signed-off-by: Scott Schreckengaust <scottschreckengaust@users.noreply.github.com> feat(construct): new construct for deployment of custom models to sagemaker (awslabs#180) * feat(construct): New construct CustomSagemakerEndpoint Fixes (New construct): Add support to deploy custom models to a sagemaker endpoint awslabs#150 * chore(construct): Update documentation * feat(construct): Add new projen task to run the code generation step (create list of available containers / jumpstart models). Can be used through projen generate-models-containers * chore(construct): Update list of available models * fix(construct): Fixes (gen-ai): JumpStartSageMakerEndpoint not working awslabs#179 : address the issue of some models from JumpStart which can't be deployed. This is due to the construct requiring model data to be provided as a compressed object, however some newer JumpStart models provide uncompressed model artifacts. Tested the fix with the reported model - whisper base 2.0.0. Deployed model and ran inference. This requires an upgrade of cdk as the update to the L1 CloudFormation resource was added recently (addition of the property https://docs.aws.amazon.com/cdk/api/v2/docs/aws-cdk-lib.aws_sagemaker.CfnModel.ModelDataSourceProperty.html) chore(deps): upgrade dependencies (awslabs#188) Upgrades project dependencies. See details in [workflow run]. [Workflow Run]: https://github.com/awslabs/generative-ai-cdk-constructs/actions/runs/7482301025 ------ *Automatically created by projen via the "upgrade-main" workflow* Signed-off-by: github-actions <github-actions@github.com> Co-authored-by: github-actions <github-actions@github.com> fix: use definition instead of schema (awslabs#187) Closes awslabs#186 Signed-off-by: Scott Schreckengaust <scottschreckengaust@users.noreply.github.com> feat: add husky with a convential commit check (awslabs#189) Signed-off-by: Scott Schreckengaust <scottschreckengaust@users.noreply.github.com> fix(doc): Update README.md (awslabs#190) Remove duplicated section from the readme Signed-off-by: Alain Krok <alkrok@amazon.com> fix: do not require husky downstream (awslabs#192) Signed-off-by: Scott Schreckengaust <scottschreckengaust@users.noreply.github.com> chore(doc): Update README_custom_sagemaker_endpoint.md (awslabs#191) Fix image not displayed Signed-off-by: Alain Krok <alkrok@amazon.com> chore(deps): upgrade dependencies (awslabs#194) Upgrades project dependencies. See details in [workflow run]. [Workflow Run]: https://github.com/awslabs/generative-ai-cdk-constructs/actions/runs/7508788235 ------ *Automatically created by projen via the "upgrade-main" workflow* Signed-off-by: github-actions <github-actions@github.com> Co-authored-by: github-actions <github-actions@github.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Describe the feature
Add a construct to enable deployment of custom models to a SageMaker endpoint (fine tuned, inf2,...)
This construct should provide the user the ability to point towards a location in Amazon S3 where the model artifacts are located, and a DLC container image
Use Case
More flexibility and customization for end users
Proposed Solution
No response
Other Information
No response
Acknowledgements
The text was updated successfully, but these errors were encountered: