Models

Models are an essential part of Neurodeploy — they are the containers for the persisted pre-trained AI models that will be deployed and will allow you to generate your predictions. On this page, we’ll dive into the different models endpoints you can use to manage models programmatically. We'll look at how to query, create, update, and delete models.

The AI models model

The AI models model contains all the information about the AI model that you pre-trained. Models can be pre-trained using Scikit-learn, TensorFlow or Pytorch. They can be persisted using H5, Joblib or Pickle.

Properties

Name
model_name
Type
string
Description
Unique identifier for the model (accross all our clients).
Name
is_public
Type
boolean
Description
Boolean to identify if a model can be accessible without using any key.
Name
library
Type
string
Description
The library used to train the AI model. Possible values : Scikit-learn, TensorFlow, Pytorch.
Name
filetype
Type
string
Description
The type of persistence used for the file. Possible values : H5, Pickle, Joblib.
Name
updated_at
Type
timestamp
Description
Timestamp of when the model was last used by the predict endpoint. user.
Name
created_at
Type
timestamp
Description
Timestamp of when the model was created.
Name
deleted_at
Type
timestamp
Description
TODO Timestamp of when the model was archived.

GET/ml-models

List all ML models

This endpoint allows you to retrieve a paginated list of all your ML models. By default, a maximum of ten models are shown per page.

Optional attributes

Name
limit
Type
integer
Description
Limit the number of models returned.
Name
deleted
Type
boolean
Description
Only show models that have deleted_at different than null.
Name
used
Type
boolean
Description
Only show models that have last_used_at different than null.

Request

GET

/ml-models

curl --location 'https://user-api.neurodeploy.com/ml-models' \
--header 'Authorization: Bearer BEARER_TOKEN'
  -d limit=10

Response

{
"models": [
    {
        "filetype": "h5",
        "library": "tensorflow",
        "created_at": "2023-05-02T07:27:06.739805",
        "is_public": false,
        "updated_at": "2023-05-10T20:39:39.325670",
        "model_name": "abc"
    },
    {
        "filetype": "h5",
        "library": "tensorflow",
        "created_at": "2023-05-11T05:10:13.916465",
        "is_public": true,
        "updated_at": "2023-05-11T05:16:49.739782",
        "model_name": "xyz"
    }
]
}

PUT/ml-models/

Update ML Model details or Create a new Model

This endpoint allows you to update some details of an ML model.

Through the sdk the deploy function also allows you to deploy a model directly by the persisted file or by the trained model.
We deploy the model that was trained above using the neuro.deploy command, providing it with a model name of our choosing. This command will send our model to the Neurodeploy infrastructure and make it accessible over the internet. With this single command, we have deployed our machine learning model!

Required attributes

Name
lib
Type
string
Description
The library used to train the AI model. Possible values : Scikit-learn, TensorFlow, Pytorch
Name
filetype
Type
string
Description
The type of persistence used for the file. Possible values : H5, Pickle, Joblib
Name
is_public
Type
boolean
Description
Boolean to identify if a model can be accessible without using any key.

Request

PUT

/ml-models/model_name

curl --location --request PUT 'https://user-api.{{domain_name}}/ml-models/{{model_name}}?lib=tensorflow&filetype=h5&is_public=true' \
--header 'Authorization: Bearer BEARER_TOKEN'

Response success

{
"model_name": "scikit_learn_open_dataset_iris_2",
"model_type": "scikit-learn",
"persistence_type": "joblib",
"uploaded_at": "2023-03-28T03:54:22+00:00"
}

Response additional actions

{
"message": "Please upload your tensorflow h5 model to complete the process.",
"url": "https://neurodeploy-staging-us-east-1.s3.amazonaws.com/",
"fields": {
    "x-amz-meta-username": "lousydropout",
    "x-amz-meta-model_name": "abc",
    "x-amz-meta-lib": "tensorflow",
    "x-amz-meta-filetype": "h5",
    "Content-Type": "model/h5",
    "key": "a52df0c8-843a-48f9-b2fb-726f3a727945",
    "AWSAccessKeyId": "ASIAWWJYOSALBH4EQXHE",
    "x-amz-security-token": "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",
    "policy": "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",
    "signature": "0FWgS0LJ2oyLFbZkmFEyAbFpTsk="
}
}

GET/ml-models/model-name

Get ML Model details

This endpoint allows you to retrieve an ML model by providing the ML model name. Refer to the list at the top of this page to see which properties are included with model objects.

Request

GET

/ml-models/model_name'

curl --location 'https://user-api.neurodeploy.com/ml-models/abc' \
--header 'Authorization: Bearer BEARER_TOKEN'

Response

{
"filetype": "h5",
"library": "tensorflow",
"created_at": "2023-05-02T07:27:06.739805",
"updated_at": "2023-05-12T16:55:25.577078",
"model_name": "abc"
}

DELETE/ml-models/{{model_name}}

Delete an ML model

This endpoint allows you to delete your ML model in Neurodeploy. Note: This will permanently delete the model and all its predictions if stores — archive it instead if you want to be able to restore it later.

Request

DELETE

/ml-models/{{model_name}}

curl --location --request DELETE 'https://user-api.neurodeploy.com/ml-models/my_model_name' \
--header 'Authorization: Bearer BEARER_TOKEN'

Response

{
  "message": "deleted model my_model_name"
}