But we need to provide our custom training image from AWS ECR and training data from AWS S3 bucket. Alternatively, we could also train the model online with the AWS SageMaker by creating a training job under the SageMaker Training section. To create a model artifact, we could simply compress this folder into a tar.gz format and upload it to a S3 bucket. The folder usually contains the following files: Every time we finish training a spaCy model, the trained model is saved into a folder. To create the SageMaker endpoint, we need to create our custom spaCy model container on the AWS ECR first and prepare our trained model artifact ready in format of tar.gz.įurthermore, to create a model artifact, we could do it either offline or online. Or perhaps connect the prediction result with another Lambda and integrate into the production workflow.ĭuring above steps, the spaCy model is hosted on AWS SageMaker and can be invoked as an endpoint at any time. In real-world scenario, we may also want to connect an AWS API GateWay for example if we want to host an interface for the model.If error occurs, we can debug on the AWS CloudWatch. Finally, this endpoint prediction will display as an execution result on AWS Lambda result tab.The endpoint job is to take the event as an input and return a prediction result from the fine-tuned spaCy model deployed on SageMaker. Then, within the AWS Lambda handler function, we invoke a SageMaker endpoint.First, raw data input can be sent as an event into AWS Lambda.In this article, I will explain my solution on how to deploy a custom spaCy model with AWS services including: It’s a very interesting topic and after a great amount of work I summarized my solution into this article hopefully it can be useful for people facing the same question. Although I have found many great tutorials on deploying a spaCy model locally with Flask / Streamlit, there are not many tutorials on how to deploy it on a larger scale, for example, how to deploy a spaCy model with AWS.
0 kommentar(er)
