Automate Amazon SageMaker Pipelines DAG creation

Creating scalable and environment friendly machine studying (ML) pipelines is essential for streamlining the event, deployment, and administration of ML fashions. On this put up, we current a framework for automating the creation of a directed acyclic graph (DAG) for Amazon SageMaker Pipelines primarily based on easy configuration information. The framework code and examples offered right here solely cowl mannequin coaching pipelines, however could be readily prolonged to batch inference pipelines as properly.

This dynamic framework makes use of configuration information to orchestrate preprocessing, coaching, analysis, and registration steps for each single-model and multi-model use instances primarily based on user-defined Python scripts, infrastructure wants (together with Amazon Virtual Private Cloud (Amazon VPC) subnets and safety teams, AWS Identity and Access Management (IAM) roles, AWS Key Management Service (AWS KMS) keys, containers registry, and occasion sorts), enter and output Amazon Simple Storage Service (Amazon S3) paths, and useful resource tags. Configuration information (YAML and JSON) permit ML practitioners to specify undifferentiated code for orchestrating coaching pipelines utilizing declarative syntax. This allows information scientists to shortly construct and iterate on ML fashions, and empowers ML engineers to run via steady integration and steady supply (CI/CD) ML pipelines sooner, reducing time to manufacturing for fashions.

Resolution overview

The proposed framework code begins by studying the configuration information. It then dynamically creates a SageMaker Pipelines DAG primarily based on the steps declared within the configuration information and the interactions and dependencies amongst steps. This orchestration framework caters to each single-model and multi-model use instances, and gives a clean move of information and processes. The next are the important thing advantages of this resolution:

• Automation – Your complete ML workflow, from information preprocessing to mannequin registry, is orchestrated with no guide intervention. This reduces the effort and time required for mannequin experimentation and operationalization.
• Reproducibility – With a predefined configuration file, information scientists and ML engineers can reproduce the whole workflow, reaching constant outcomes throughout a number of runs and environments.
• Scalability – Amazon SageMaker is used all through the pipeline, enabling ML practitioners to course of massive datasets and practice advanced fashions with out infrastructure issues.
• Flexibility – The framework is versatile and might accommodate a variety of ML use instances, ML frameworks (equivalent to XGBoost and TensorFlow), multi-model coaching, and multi-step coaching. Each step of the coaching DAG could be custom-made through the configuration file.
• Mannequin governance – The Amazon SageMaker Model Registry integration permits for monitoring mannequin variations, and due to this fact selling them to manufacturing with confidence.

The next structure diagram depicts how you should use the proposed framework throughout each experimentation and operationalization of ML fashions. Throughout experimentation, you’ll be able to clone the framework code repository supplied on this put up and your project-specific supply code repositories into Amazon SageMaker Studio, and set your digital setting (detailed later on this put up). You may then iterate on preprocessing, coaching, and analysis scripts, in addition to configuration decisions. To create and run a SageMaker Pipelines coaching DAG, you’ll be able to name the framework’s entry level, which can learn all of the configuration information, create the mandatory steps, and orchestrate them primarily based on the required step ordering and dependencies.

Throughout operationalization, the CI pipeline clones the framework code repository and project-specific coaching repositories into an AWS CodeBuild job, the place the framework’s entry level script known as to create or replace the SageMaker Pipelines coaching DAG, after which run it.

Repository construction

The GitHub repository accommodates the next directories and information:

• /framework/conf/ – This listing accommodates a configuration file that’s used to set frequent variables throughout all modeling models equivalent to subnets, safety teams, and IAM position on the runtime. A modeling unit is a sequence of as much as six steps for coaching an ML mannequin.
• /framework/createmodel/ – This listing accommodates a Python script that creates a SageMaker model object primarily based on mannequin artifacts from a SageMaker Pipelines training step. The mannequin object is later utilized in a SageMaker batch transform job for evaluating mannequin efficiency on a take a look at set.
• /framework/modelmetrics/ – This listing accommodates a Python script that creates an Amazon SageMaker Processing job for producing a mannequin metrics JSON report for a skilled mannequin primarily based on outcomes of a SageMaker batch remodel job carried out on take a look at information.
• /framework/pipeline/ – This listing accommodates Python scripts that use Python courses outlined in different framework directories to create or replace a SageMaker Pipelines DAG primarily based on the required configurations. The model_unit.py script is utilized by pipeline_service.py to create a number of modeling models. Every modeling unit is a sequence of as much as six steps for coaching an ML mannequin: course of, practice, create mannequin, remodel, metrics, and register mannequin. Configurations for every modeling unit must be specified within the mannequin’s respective repository. The pipeline_service.py additionally units dependencies amongst SageMaker Pipelines steps (how steps inside and throughout modeling models are sequenced or chained) primarily based on the sagemakerPipeline part, which must be outlined within the configuration file of one of many mannequin repositories (the anchor mannequin). This lets you override default dependencies inferred by SageMaker Pipelines. We talk about the configuration file construction later on this put up.
• /framework/processing/ – This listing accommodates a Python script that creates a SageMaker Processing job primarily based on the required Docker picture and entry level script.
• /framework/registermodel/ – This listing accommodates a Python script for registering a skilled mannequin together with its calculated metrics in SageMaker Mannequin Registry.
• /framework/coaching/ – This listing accommodates a Python script that creates a SageMaker coaching job.
• /framework/remodel/ – This listing accommodates a Python script that creates a SageMaker batch remodel job. Within the context of mannequin coaching, that is used to calculate the efficiency metric of a skilled mannequin on take a look at information.
• /framework/utilities/ – This listing accommodates utility scripts for studying and becoming a member of configuration information, in addition to logging.
• /framework_entrypoint.py – This file is the entry level of the framework code. It calls a operate outlined within the /framework/pipeline/ listing to create or replace a SageMaker Pipelines DAG and run it.
• /examples/ – This listing accommodates a number of examples of how you should use this automation framework to create easy and complicated coaching DAGs.
• /env.env – This file permits you to set frequent variables equivalent to subnets, safety teams, and IAM position as setting variables.
• /necessities.txt – This file specifies Python libraries which can be required for the framework code.

Stipulations

It’s best to have the next conditions earlier than deploying this resolution:

• An AWS account
• SageMaker Studio
• A SageMaker position with Amazon S3 learn/write and AWS KMS encrypt/decrypt permissions
• An S3 bucket for storing information, scripts, and mannequin artifacts
• Optionally, the AWS Command Line Interface (AWS CLI)
• Python3 (Python 3.7 or larger) and the next Python packages:
• Extra Python packages utilized in your customized scripts

Deploy the answer

Full the next steps to deploy the answer:

1. Set up your mannequin coaching repository in line with the next construction:

   <MODEL-DIR-REPO>
    .
   ├── <MODEL-DIR>
   |    ├── conf
   |    |   └── conf.yaml
   |    └── scripts
   |        ├── preprocess.py
   |        ├── practice.py
   |        ├── remodel.py
   |        └── consider.py
   └── README.md
   

2. Clone the framework code and your mannequin supply code from the Git repositories:

• • Clone dynamic-sagemaker-pipelines-framework repo right into a coaching listing. Within the following code, we assume the coaching listing known as aws-train:

    git clone https://github.com/aws-samples/dynamic-sagemaker-pipelines-framework.git aws-train

  • Clone the mannequin supply code underneath the identical listing. For multi-model coaching, repeat this step for as many fashions as you could practice.

    git clone https:<MODEL-DIR-REPO>.git aws-train

For single-model coaching, your listing ought to seem like the next:

<aws-train>  
.  
├── framework
└── <MODEL-DIR>

For multi-model coaching, your listing ought to seem like the next:

<aws-train>  
.  
├── framework
└── <MODEL-DIR-1>
└── <MODEL-DIR-2>
└── <MODEL-DIR-3>

3. Arrange the next setting variables. Asterisks point out setting variables which can be required; the remainder are non-compulsory.

For single-model use instances, SMP_MODEL_CONFIGPATH might be <MODEL-DIR>/conf/conf.yaml. For multi-model use instances, SMP_MODEL_CONFIGPATH might be */conf/conf.yaml, which lets you discover all conf.yaml information utilizing Python’s glob module and mix them to kind a worldwide configuration file. Throughout experimentation (native testing), you’ll be able to specify setting variables contained in the env.env file after which export them by operating the next command in your terminal:

Notice that the values of setting variables in env.env must be positioned inside citation marks (for instance, SMP_REGION="us-east-1"). Throughout operationalization, these setting variables must be set by the CI pipeline.

4. Create and activate a digital setting by operating the next instructions:

   python -m venv .venv
   
   supply .venv/bin/activate

5. Set up the required Python packages by operating the next command:

   pip set up -r necessities.txt

6. Edit your mannequin coaching conf.yaml information. We talk about the configuration file construction within the subsequent part.
7. From the terminal, name the framework’s entry level to create or replace and run the SageMaker Pipeline coaching DAG:

   python framework/framework_entrypoint.py

8. View and debug the SageMaker Pipelines run on the Pipelines tab of the SageMaker Studio UI.

Configuration file construction

There are two sorts of configuration information within the proposed resolution: framework configuration and mannequin configuration. On this part, we describe every intimately.

Framework configuration

The /framework/conf/conf.yaml file units the variables which can be frequent throughout all modeling models. This consists of SMP_S3BUCKETNAME, SMP_ROLE, SMP_MODEL_CONFIGPATH, SMP_SUBNETS, SMP_SECURITYGROUPS, and SMP_MODELNAME. Confer with Step 3 of deployment directions for descriptions of those variables and set them through setting variables.

Mannequin configuration

For every mannequin within the undertaking, we have to specify the next within the <MODEL-DIR>/conf/conf.yaml file (asterisks point out required sections; the remainder are non-compulsory):

• /conf/fashions* – On this part, you’ll be able to configure a number of modeling models. When the framework code is run, it is going to mechanically learn all configuration information throughout runtime and append them to the config tree. Theoretically, you’ll be able to specify all modeling models in the identical conf.yaml file, but it surely’s really helpful to specify every modeling unit configuration in its respective listing or Git repository to attenuate errors. The models are as follows:
  • {model-name}* – The title of the mannequin.
  • source_directory* – A standard source_dir path to make use of for all steps inside the modeling unit.
  • preprocess – This part specifies preprocessing parameters.
  • practice* – This part specifies coaching job parameters.
  • remodel* – This part specifies SageMaker Remodel job parameters for making predictions on the take a look at information.
  • consider – This part specifies SageMaker Processing job parameters for producing a mannequin metrics JSON report for the skilled mannequin.
  • registry* – This part specifies parameters for registering the skilled mannequin in SageMaker Mannequin Registry.
• /conf/sagemakerPipeline* – This part defines the SageMaker Pipelines move, together with dependencies amongst steps. For single-model use instances, this part is outlined on the finish of the configuration file. For multi-model use instances, the sagemakerPipeline part solely must be outlined within the configuration file of one of many fashions (any of the fashions). We consult with this mannequin because the anchor mannequin. The parameters are as follows:
  • pipelineName* – Identify of the SageMaker pipeline.
  • fashions* – Nested checklist of modeling models:
    • {model-name}* – Mannequin identifier, which ought to match a {model-name} identifier within the /conf/fashions part.
      • steps* –
        • step_name* – Step title to be displayed within the SageMaker Pipelines DAG.
        • step_class* – (Union[Processing, Training, CreateModel, Transform, Metrics, RegisterModel])
        • step_type* – This parameter is simply required for preprocessing steps, for which it must be set to preprocess. That is wanted to tell apart preprocess and consider steps, each of which have a step_class of Processing.
        • enable_cache – ([Union[True, False]]). This means whether or not to allow SageMaker Pipelines caching for this step.
        • chain_input_source_step – ([list[step_name]]). You need to use this to set the channel outputs of one other step as enter to this step.
        • chain_input_additional_prefix – That is solely allowed for steps of the Remodel step_class, and can be utilized along with chain_input_source_step parameter to pinpoint the file that must be used because the enter to the remodel step.
  • dependencies – This part specifies the sequence during which the SageMaker Pipelines steps must be run. We’ve got tailored the Apache Airflow notation for this part (for instance, {step_name} >> {step_name}). If this part is left clean, express dependencies specified by the chain_input_source_step parameter or implicit dependencies outline the SageMaker Pipelines DAG move.

Notice that we suggest having one coaching step per modeling unit. If a number of coaching steps are outlined for a modeling unit, the next steps implicitly take the final coaching step to create the mannequin object, calculate metrics, and register the mannequin. If you could practice a number of fashions, it’s really helpful to create a number of modeling models.

Examples

On this part, we exhibit three examples of ML mannequin coaching DAGs created utilizing the offered framework.

Single-model coaching: LightGBM

It is a single-model instance for a classification use case the place we use LightGBM in script mode on SageMaker. The dataset consists of categorical and numerical variables to foretell the binary label Income (to foretell if the topic makes a purchase order or not). The preprocessing script is used to mannequin the information for coaching and testing after which stage it in an S3 bucket. The S3 paths are then supplied to the training step within the configuration file.

When the coaching step runs, SageMaker masses the file on the container at /decide/ml/enter/information/{channelName}/, accessible through the setting variable SM_CHANNEL_{channelName} on the container (channelName= ‘train’ or ‘test’).The training script does the next:

1. Load the information domestically from native container paths utilizing the NumPy load module.
2. Set hyperparameters for the coaching algorithm.
3. Save the skilled mannequin on the native container path /decide/ml/mannequin/.

SageMaker takes the content material underneath /decide/ml/mannequin/ to create a tarball that’s used to deploy the mannequin to SageMaker for internet hosting.

The remodel step takes as enter the staged test file as input and the skilled mannequin to make predictions on the skilled mannequin. The output of the remodel step is chained to the metrics step to guage the mannequin in opposition to the ground truth, which is explicitly provided to the metrics step. Lastly, the output of the metrics step is implicitly chained to the register step to register the mannequin in SageMaker Mannequin Registry with details about the mannequin’s efficiency produced within the metrics step. The next determine exhibits a visible illustration of the coaching DAG. You may consult with the scripts and configuration file for this instance within the GitHub repo.

Single-model coaching: LLM fine-tuning

That is one other single-model coaching instance, the place we orchestrate fine-tuning of a Falcon-40B massive language mannequin (LLM) from Hugging Face Hub for a textual content summarization use case. The preprocessing script masses the samsum dataset from Hugging Face, masses the tokenizer for the mannequin, and processes the practice/take a look at information splits for fine-tuning the mannequin on this area information within the falcon-text-summarization-preprocess step.

The output is chained to the falcon-text-summarization-tuning step, the place the training script masses the Falcon-40B LLM from Hugging Face Hub and begins accelerated fine-tuning utilizing LoRA on the practice break up. The mannequin is evaluated in the identical step after fine-tuning, which gatekeeps the analysis loss to fail the falcon-text-summarization-tuning step, which causes the SageMaker pipeline to cease earlier than it is ready to register the fine-tuned mannequin. In any other case, the falcon-text-summarization-tuning step runs efficiently and the mannequin is registered in SageMaker Mannequin Registry. The next determine exhibits a visible illustration of the LLM fine-tuning DAG. The scripts and configuration file for this instance can be found within the GitHub repo.

Multi-model coaching

It is a multi-model coaching instance the place a principal part evaluation (PCA) mannequin is skilled for dimensionality discount, and a TensorFlow Multilayer Perceptron mannequin is skilled for California Housing Price prediction. The TensorFlow mannequin’s preprocessing step makes use of a skilled PCA mannequin to cut back dimensionality of its coaching information. We add a dependency within the configuration to make sure the TensorFlow mannequin is registered after PCA mannequin registration. The next determine exhibits a visible illustration of the multi-model coaching DAG instance. The scripts and configuration information for this instance can be found within the GitHub repo.

Clear up

Full the next steps to wash up your assets:

1. Use the AWS CLI to list and remove any remaining pipelines which can be created by the Python scripts.
2. Optionally, delete different AWS assets such because the S3 bucket or IAM position created exterior SageMaker Pipelines.

Conclusion

On this put up, we offered a framework for automating SageMaker Pipelines DAG creation primarily based on configuration information. The proposed framework provides a forward-looking resolution to the problem of orchestrating advanced ML workloads. By utilizing a configuration file, SageMaker Pipelines gives the pliability to construct orchestration with minimal code, so you’ll be able to streamline the method of making and managing each single-model and multi-model pipelines. This method not solely saves time and assets, but additionally promotes MLOps greatest practices, contributing to the general success of ML initiatives. For extra details about implementation particulars, evaluation the GitHub repo.

Concerning the Authors

Luis Felipe Yepez Barrios, is a Machine Studying Engineer with AWS Skilled Providers, centered on scalable distributed methods and automation tooling to expedite scientific innovation within the area of Machine Studying (ML). Moreover, he assists enterprise shoppers in optimizing their machine studying options via AWS providers.

Jinzhao Feng, is a Machine Studying Engineer at AWS Skilled Providers. He focuses on architecting and implementing massive scale Generative AI and classical ML pipeline options. He’s specialised in FMOps, LLMOps and distributed coaching.

Harsh Asnani, is a Machine Studying Engineer at AWS. His Background is in Utilized Knowledge Science with a concentrate on operationalizing Machine Studying workloads within the cloud at scale.

Hasan Shojaei, is a Sr. Knowledge Scientist with AWS Skilled Providers, the place he helps prospects throughout completely different industries clear up their enterprise challenges via the usage of large information, machine studying, and cloud applied sciences. Previous to this position, Hasan led a number of initiatives to develop novel physics-based and data-driven modeling methods for high power firms. Outdoors of labor, Hasan is keen about books, climbing, pictures, and historical past.

Alec Jenab, is a Machine Studying Engineer who focuses on creating and operationalizing machine studying options at scale for enterprise prospects. Alec is keen about bringing progressive options to market, particularly in areas the place machine studying can meaningfully enhance finish person expertise. Outdoors of labor, he enjoys taking part in basketball, snowboarding, and discovering hidden gems in San Francisco.