Specifications for Editing a Model Configuration File
A model developer needs to edit a configuration file config.json when publishing a model. The model configuration file describes the model usage, computing framework, precision, inference code dependency package, and model API.
Configuration File Format
The configuration file is in JSON format. Table 1 describes the parameters.
Parameter |
Mandatory |
Data Type |
Description |
|---|---|---|---|
model_algorithm |
Yes |
String |
Model algorithm, which is set by the model developer to help model users understand the usage of the model. The value must start with a letter and contain no more than 36 characters. Chinese characters and special characters (&!'\"<>=) are not allowed. Common model algorithms include image_classification (image classification), object_detection (object detection), and predict_analysis (prediction analysis). |
model_type |
Yes |
String |
Model AI engine, which indicates the computing framework used by a model. Common AI engines and Image are supported.
|
runtime |
No |
String |
Model runtime environment. Python2.7 is used by default The value of runtime depends on the value of model_type. If model_type is set to Image, you do not need to set runtime. If model_type is set to another mainstream framework, select the engine and runtime environment. For details about the supported running environments, see Supported AI Engines for ModelArts Inference. If your model must run on specified CPUs or GPUs, select the CPUs or GPUs based on the runtime suffix. If the runtime does not contain the CPU or GPU information, check the runtime description in Supported AI Engines for ModelArts Inference. |
metrics |
No |
Object |
Model precision information, including the average value, recall rate, precision, and accuracy. For details about the metrics object structure, see Table 2. The result is displayed in the model precision area on the AI application details page. |
apis |
No |
api array |
Format of the requests received and returned by a model. The value is structure data. It is the RESTful API array provided by a model. For details about the API data structure, see Table 3. For details about the code example, see Code Example of apis Parameters.
|
dependencies |
No |
dependency array |
Package on which the model inference code depends, which is structure data. Model developers need to provide the package name, installation mode, and version constraints. Only the pip installation mode is supported. Table 6 describes the dependency array. If the model package does not contain the customize_service.py file, you do not need to set this parameter. Dependency packages cannot be installed for custom image models.
NOTE:
The dependencies parameter supports multiple dependency structure arrays in list format and applies to scenarios where the default installation packages have dependency relationships. Packages on the top are installed first. The wheel package on premises can be used for installation. (The wheel package must be stored in the same directory as the model file). For details, see How Do I Edit the Installation Package Dependency Parameters in a Model Configuration File When Importing a Model? |
health |
No |
health data structure |
Configuration of an image health interface. This parameter is mandatory only when model_type is set to Image. If services cannot be interrupted during a rolling upgrade, a health check API must be provided for ModelArts to call. For details about the health data structure, see Table 8. |
Parameter |
Mandatory |
Data Type |
Description |
|---|---|---|---|
f1 |
No |
Number |
F1 score. The value is rounded to 17 decimal places. |
recall |
No |
Number |
Recall rate. The value is rounded to 17 decimal places. |
precision |
No |
Number |
Precision. The value is rounded to 17 decimal places. |
accuracy |
No |
Number |
Accuracy. The value is rounded to 17 decimal places. |
Parameter |
Mandatory |
Data Type |
Description |
|---|---|---|---|
url |
No |
String |
Request path. The default value is a slash (/). For a custom image model (model_type is Image), set this parameter to the actual request path exposed in the image. For a non-custom image model (model_type is not Image), the URL can only be /. |
method |
No |
String |
Request method. The default value is POST. |
request |
No |
Object |
Request body. For details, see Table 4. |
response |
No |
Object |
Response body. For details, see Table 5. |
Parameter |
Mandatory |
Data Type |
Description |
|---|---|---|---|
Content-type |
No for real-time services Yes for batch services |
String |
Data is sent in a specified content format. The default value is application/json. The options are as follows:
NOTE:
For machine learning models, only application/json is supported. |
data |
No for real-time services Yes for batch services |
String |
The request body is described in JSON schema. For details about the parameter description, see the official guide. |
Parameter |
Mandatory |
Data Type |
Description |
|---|---|---|---|
Content-type |
No for real-time services Yes for batch services |
String |
Data is sent in a specified content format. The default value is application/json.
NOTE:
For machine learning models, only application/json is supported. |
data |
No for real-time services Yes for batch services |
String |
The response body is described in JSON schema. For details about the parameter description, see the official guide. |
Parameter |
Mandatory |
Data Type |
Description |
|---|---|---|---|
installer |
Yes |
String |
Installation method. Only pip is supported. |
packages |
Yes |
package array |
Dependency package collection. For details about the package structure array, see Table 7. |
Parameter |
Mandatory |
Type |
Description |
|---|---|---|---|
package_name |
Yes |
String |
Dependency package name. Chinese characters and special characters (&!'"<>=) are not allowed. |
package_version |
No |
String |
Dependency package version. If the dependency package does not rely on package versions, leave this field blank. Chinese characters and special characters (&!'"<>=) are not allowed. |
restraint |
No |
String |
Version restriction. This parameter is mandatory only when package_version is configured. Possible values are EXACT, ATLEAST, and ATMOST.
|
Parameter |
Mandatory |
Type |
Description |
|---|---|---|---|
check_method |
Yes |
String |
Health check method. The value can be HTTP or EXEC.
|
command |
No |
String |
Health check command. This parameter is mandatory when check_method is set to EXEC. |
url |
No |
String |
Request URL of a health check API. This parameter is mandatory when check_method is set to HTTP. |
protocol |
No |
String |
Request protocol of a health check API. The default value is http. This parameter is mandatory when check_method is set to HTTP. |
initial_delay_seconds |
No |
String |
Delay for initializing the health check. |
timeout_seconds |
No |
String |
Health check timeout. |
period_seconds |
Yes |
String |
Health check period, in seconds. Enter an integer greater than 0 and no more than 2147483647. |
failure_threshold |
Yes |
String |
Maximum number of health check failures. Enter an integer greater than 0 and no more than 2147483647. |
Code Example of apis Parameters
[{
"url": "/",
"method": "post",
"request": {
"Content-type": "multipart/form-data",
"data": {
"type": "object",
"properties": {
"images": {
"type": "file"
}
}
}
},
"response": {
"Content-type": "applicaton/json",
"data": {
"type": "object",
"properties": {
"mnist_result": {
"type": "array",
"item": [
{
"type": "string"
}
]
}
}
}
}
}]
Example of the Object Detection Model Configuration File
The following code uses the TensorFlow engine as an example. You can modify the model_type parameter based on the actual engine type.
- Model input
Value: image files
- Model output
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
{ "detection_classes": [ "face", "arm" ], "detection_boxes": [ [ 33.6, 42.6, 104.5, 203.4 ], [ 103.1, 92.8, 765.6, 945.7 ] ], "detection_scores": [0.99, 0.73] }
- Configuration file
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70
{ "model_type": "TensorFlow", "model_algorithm": "object_detection", "metrics": { "f1": 0.345294, "accuracy": 0.462963, "precision": 0.338977, "recall": 0.351852 }, "apis": [{ "url": "/", "method": "post", "request": { "Content-type": "multipart/form-data", "data": { "type": "object", "properties": { "images": { "type": "file" } } } }, "response": { "Content-type": "application/json", "data": { "type": "object", "properties": { "detection_classes": { "type": "array", "items": [{ "type": "string" }] }, "detection_boxes": { "type": "array", "items": [{ "type": "array", "minItems": 4, "maxItems": 4, "items": [{ "type": "number" }] }] }, "detection_scores": { "type": "array", "items": [{ "type": "number" }] } } } } }], "dependencies": [{ "installer": "pip", "packages": [{ "restraint": "EXACT", "package_version": "1.15.0", "package_name": "numpy" }, { "restraint": "EXACT", "package_version": "5.2.0", "package_name": "Pillow" } ] }] }
Example of the Image Classification Model Configuration File
The following code uses the TensorFlow engine as an example. You can modify the model_type parameter based on the actual engine type.
- Model input
Value: image files
- Model output
1 2 3 4 5 6 7
{ "predicted_label": "flower", "scores": [ ["rose", 0.99], ["begonia", 0.01] ] }
- Configuration file
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66
{ "model_type": "TensorFlow", "model_algorithm": "image_classification", "metrics": { "f1": 0.345294, "accuracy": 0.462963, "precision": 0.338977, "recall": 0.351852 }, "apis": [{ "url": "/", "method": "post", "request": { "Content-type": "multipart/form-data", "data": { "type": "object", "properties": { "images": { "type": "file" } } } }, "response": { "Content-type": "application/json", "data": { "type": "object", "properties": { "predicted_label": { "type": "string" }, "scores": { "type": "array", "items": [{ "type": "array", "minItems": 2, "maxItems": 2, "items": [ { "type": "string" }, { "type": "number" } ] }] } } } } }], "dependencies": [{ "installer": "pip", "packages": [{ "restraint": "ATLEAST", "package_version": "1.15.0", "package_name": "numpy" }, { "restraint": "", "package_version": "", "package_name": "Pillow" } ] }] }
The following code uses the MindSpore engine as an example. You can modify the model_type parameter based on the type of the engine you use.
- Model input
Value: image files
- Model output
1"[[-2.404526 -3.0476532 -1.9888215 0.45013925 -1.7018927 0.40332815\n -7.1861157 11.290332 -1.5861531 5.7887416 ]]"
- Configuration file
{ "model_algorithm": "image_classification", "model_type": "MindSpore", "metrics": { "f1": 0.124555, "recall": 0.171875, "precision": 0.0023493892851938493, "accuracy": 0.00746268656716417 }, "apis": [{ "url": "/", "method": "post", "request": { "Content-type": "multipart/form-data", "data": { "type": "object", "properties": { "images": { "type": "file" } } } }, "response": { "Content-type": "applicaton/json", "data": { "type": "object", "properties": { "mnist_result": { "type": "array", "item": [{ "type": "string" }] } } } } } ], "dependencies": [] }
Example of the Predictive Analytics Model Configuration File
The following code uses the TensorFlow engine as an example. You can modify the model_type parameter based on the actual engine type.
- Model input
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
{ "data": { "req_data": [ { "buying_price": "high", "maint_price": "high", "doors": "2", "persons": "2", "lug_boot": "small", "safety": "low", "acceptability": "acc" }, { "buying_price": "high", "maint_price": "high", "doors": "2", "persons": "2", "lug_boot": "small", "safety": "low", "acceptability": "acc" } ] } }
- Model output
1 2 3 4 5 6 7 8 9 10 11 12
{ "data": { "resp_data": [ { "predict_result": "unacc" }, { "predict_result": "unacc" } ] } }
- Configuration file
In the code, the data parameter in the request and response structures is described in JSON Schema. The content in data and properties corresponds to the model input and output.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77
{ "model_type": "TensorFlow", "model_algorithm": "predict_analysis", "metrics": { "f1": 0.345294, "accuracy": 0.462963, "precision": 0.338977, "recall": 0.351852 }, "apis": [ { "url": "/", "method": "post", "request": { "Content-type": "application/json", "data": { "type": "object", "properties": { "data": { "type": "object", "properties": { "req_data": { "items": [ { "type": "object", "properties": {} } ], "type": "array" } } } } } }, "response": { "Content-type": "application/json", "data": { "type": "object", "properties": { "data": { "type": "object", "properties": { "resp_data": { "type": "array", "items": [ { "type": "object", "properties": {} } ] } } } } } } } ], "dependencies": [ { "installer": "pip", "packages": [ { "restraint": "EXACT", "package_version": "1.15.0", "package_name": "numpy" }, { "restraint": "EXACT", "package_version": "5.2.0", "package_name": "Pillow" } ] } ] }
Example of the Custom Image Model Configuration File
The model input and output are similar to those in Example of the Object Detection Model Configuration File.
- If the input is an image, the request example is as follows.
In the example, a model prediction request containing the parameter images with the parameter type of file is received. For this example, the file upload button is displayed on the inference page, and the inference is performed in file format.
1 2 3 4 5 6 7 8 9 10 11
{ "Content-type": "multipart/form-data", "data": { "type": "object", "properties": { "images": { "type": "file" } } } } - If the input is JSON data, the request example is as follows.
In this example, the model prediction JSON request body is received. In the request, there is only one prediction request containing the parameter input with the parameter type of string. On the inference page, a text box is displayed for you to enter the prediction request.
1 2 3 4 5 6 7 8 9 10 11
{ "Content-type": "application/json", "data": { "type": "object", "properties": { "input": { "type": "string" } } } }
A complete request example is as follows:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 |
{
"model_algorithm": "image_classification",
"model_type": "Image",
"metrics": {
"f1": 0.345294,
"accuracy": 0.462963,
"precision": 0.338977,
"recall": 0.351852
},
"apis": [{
"url": "/",
"method": "post",
"request": {
"Content-type": "multipart/form-data",
"data": {
"type": "object",
"properties": {
"images": {
"type": "file"
}
}
}
},
"response": {
"Content-type": "application/json",
"data": {
"type": "object",
"required": [
"predicted_label",
"scores"
],
"properties": {
"predicted_label": {
"type": "string"
},
"scores": {
"type": "array",
"items": [{
"type": "array",
"minItems": 2,
"maxItems": 2,
"items": [{
"type": "string"
},
{
"type": "number"
}
]
}]
}
}
}
}
}]
}
|
Example of the Machine Learning Model Configuration File
The following uses XGBoost as an example:
- Model input
{
"req_data": [
{
"sepal_length": 5,
"sepal_width": 3.3,
"petal_length": 1.4,
"petal_width": 0.2
},
{
"sepal_length": 5,
"sepal_width": 2,
"petal_length": 3.5,
"petal_width": 1
},
{
"sepal_length": 6,
"sepal_width": 2.2,
"petal_length": 5,
"petal_width": 1.5
}
]
}
- Model output
{
"resp_data": [
{
"predict_result": "Iris-setosa"
},
{
"predict_result": "Iris-versicolor"
}
]
}
- Configuration file
{
"model_type": "XGBoost",
"model_algorithm": "xgboost_iris_test",
"runtime": "python2.7",
"metrics": {
"f1": 0.345294,
"accuracy": 0.462963,
"precision": 0.338977,
"recall": 0.351852
},
"apis": [
{
"url": "/",
"method": "post",
"request": {
"Content-type": "application/json",
"data": {
"type": "object",
"properties": {
"req_data": {
"items": [
{
"type": "object",
"properties": {}
}
],
"type": "array"
}
}
}
},
"response": {
"Content-type": "applicaton/json",
"data": {
"type": "object",
"properties": {
"resp_data": {
"type": "array",
"items": [
{
"type": "object",
"properties": {
"predict_result": {}
}
}
]
}
}
}
}
}
]
}
Example of a Model Configuration File Using a Custom Dependency Package
The following example defines the NumPy 1.16.4 dependency environment.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 |
{
"model_algorithm": "image_classification",
"model_type": "TensorFlow",
"runtime": "python3.6",
"apis": [
{
"url": "/",
"method": "post",
"request": {
"Content-type": "multipart/form-data",
"data": {
"type": "object",
"properties": {
"images": {
"type": "file"
}
}
}
},
"response": {
"Content-type": "applicaton/json",
"data": {
"type": "object",
"properties": {
"mnist_result": {
"type": "array",
"item": [
{
"type": "string"
}
]
}
}
}
}
}
],
"metrics": {
"f1": 0.124555,
"recall": 0.171875,
"precision": 0.00234938928519385,
"accuracy": 0.00746268656716417
},
"dependencies": [
{
"installer": "pip",
"packages": [
{
"restraint": "EXACT",
"package_version": "1.16.4",
"package_name": "numpy"
}
]
}
]
}
|
Feedback
Was this page helpful?
Provide feedbackThank you very much for your feedback. We will continue working to improve the documentation.See the reply and handling status in My Cloud VOC.
For any further questions, feel free to contact us through the chatbot.
Chatbot
