74 KiB

Raw Blame History

title

language_tabs

language_clients

toc_footers

includes

headingLevel

Stable Diffusion AWS extension API v2023-07-05

python
Python

javascript
Javascript

python

javascript

Stable Diffusion 亚马逊云科技插件 API

概述

本文档描述了Stable Diffusion 亚马逊云科技插件解决方案的所有 API。此解决方案包括两个部分，一部分是Stable Diffusion WebUI 扩展，它是基于 Gradio 的客户端，提供用户友好的界面，另一部分是称为中间件的资源部署在 AWS 云上，中间件提供了多个 API 接口，允许Stable Diffusion 亚马逊云科技插件客户端与 AWS 云上的服务进行交互，如 Sagemaker 和 S3，以执行模型更新/训练和推理操作。

为了支持不使用Stable Diffusion WebUI的用户，我们提供了本文档，列出了所有的 API 接口，以帮助用户了解如何调用 API 方法进行训练或推理。

在用户部署解决方案中间件 CloudFormation 后，用户可以在主堆栈的输出部分获取 API URL 和令牌。请参考此文档。

URL示例:

https://{api_id}.execute-api.{region}.amazonaws.com/prod

认证

API Key (api_key)
- API 调用时您需要在标头的x-api-key参数中配置此值。
Authorization
- 在 headers 中使用 Bearer Token

用户场景

1. 部署新的终端节点

调用 /endpoints 来创建一个新的 Sagemaker 终端节点，您需要指定两个用于创建的参数，一个是 instance_type，可选值为 "ml.g4dn.2xlarge"、"ml.g4dn.4xlarge"、"ml.g4dn.8xlarge"、"ml.g4dn.12xlarge"，另一个是 initial_instance_count，可选值为 1|2|3|4|5|6。

在调用 /endpoints 后，您需要调用 /endpoints 来列出所有终端节点的状态。通常需要超过10分钟时间才能使新的 Sagemaker 终端节点更改为 "InService" 状态。Sagemaker 终端节点只能在 "InService" 状态下用于推理。

如果终端节点处于失败状态，您可以调用 /inference/get-endpoint-deployment-job 并传递参数 jobID，响应将显示终端节点部署失败的原因，通常是由于 AWS 帐户配额限制引起的。

sequence digram raw

title Create a Sagemaker Endpoint

Client->Middleware:Call /endpoints Middleware->Middleware: Start a workflow to configure sagemaker endpoint \n based on uer request configuration Client->Middleware:Call /endpoints \n to list all the endpoint creation job list Client->Middleware:Call /inference/get-endpoint-deployment-job \n to check whether Sagemaker endpoint is in \n 'InService' state.

2. 上传模型

3. 训练模型

4. 进行推理

在 Sagemaker 终端节点处于 "InService" 状态之后，您可以调用 /inference/v2 来进行文本到图像或图像到图像的推理。您需要在请求的 POST 主体中的 "sagemaker_endpoint" 参数中指定终端节点的名称。其他所需参数位于 /inference/v2 中。

/inference/v2 将向客户端返回以下 JSON 结构：

{
  "inference_id": "XXXXXXX",
  "status": "inprogress | failed",
  "endpoint_name": "NAME_OF_ENDPOINT",
  "output_path": "path_of_prediction_output"
}

然后，客户端可以调用 /inference/get-inference-job，并将 inference_id 作为参数来查询推理作业的状态。如果推理作业成功完成（状态为 "succeed"），客户端可以使用 /inference/get-inference-job-image-output 来获取所有推理结果图像，这些图像将以 S3 预签名 URL 列表的形式返回，以便客户端下载。以下是一个示例，展示了 get-inference-job-image-output 的结果：

[
  "https://stable-diffusion-aws-extension-aigcbucketa457cb49-1tlr2pqwkosg3.s3.amazonaws.com/out/1f9679f3-25b8-4c44-8345-0a845da30094/result/image_0.jpg"
]

此外，客户端还可以调用 /inference/get-inference-job-param-output 来获取所有推理参数，/inference/get-inference-job-param-output 的响应是一个包含参数的 JSON 格式的 S3 预签名 URL，以下是一个响应示例：

[
  "https://stable-diffusion-aws-extension-aigcbucketa457cb49-1tlr2pqwkosg3.s3.amazonaws.com/out/1f9679f3-25b8-4c44-8345-0a845da30094/result/1f9679f3-25b8-4c44-8345-0a845da30094_param.json"
]

sequence digram raw

title Do Inference

Client->Middleware:Call /inference/v2 Middleware->Middleware: Start an async inference job \n on configure sagemaker endpoint \n based on uer request configuration Middleware->Client: return inference_id Client->Middleware:Call /inference/get-inference-job \n to query the inference job status Middleware->Client: return inference_id and the job status(inprocess | succeed | failure)

abox over Client: If the inference job is succeed, \n call /inference/get-inference-job-image-output and \n /inference/get-inference-job-param-output to get the \n inference result Client->Middleware:Call /inference/get-inference-job-image-output \n to get all inference result images.

Middleware->Client: return the inference result images in presigned url format

Client->Middleware:Call /inference/get-inference-job-param-output \n to get inference parameters.

Middleware->Client: return the inference parameter in presigned url format

# API 列表

序号	HTTP 方法	API 名称	描述
1	GET	/inference/test-connection	测试客户端是否可以连接到 API 并检查 API_TOKEN 是否正确
2	GET	/inference/list-inference-jobs	列出所有推理作业。
3	GET	/inference/get-inference-job	获取特定推理作业的详细信息。
4	GET	/inference/get-inference-job-image-output	获取特定推理作业的图像输出。
5	GET	/inference/get-inference-job-param-output	获取特定推理作业的参数输出。
6	POST	/inference/v2	使用默认参数运行 SageMaker 推理
7	POST	/endpoints	部署 SageMaker 终端节点。
8	DELETE	/endpoints	删除 SageMaker 终端节点。
9	GET	/endpoints	列出所有终端节点部署作业。
10	GET	/inference/get-endpoint-deployment-job	获取特定终端节点部署作业。
11	GET	/inference/generate-s3-presigned-url-for-uploading	生成用于上传的 S3 预签名 URL。
12	GET	/inference/get-texual-inversion-list	获取文本反演列表。
13	GET	/inference/get-lora-list	获取 LoRa 列表。
14	GET	/inference/get-hypernetwork-list	获取超网络列表。
15	GET	/inference/get-controlnet-model-list	获取 ControlNet 模型列表。
16	POST	/inference/run-model-merge	运行模型合并。
17	POST	/model	创建新模型。
18	PUT	/model	上传模型文件。
19	GET	/models	列出所有模型。
20	GET	/checkpoint	获取检查点。
21	PUT	/checkpoint	更新检查点。
22	GET	/checkpoints	列出所有检查点。
23	PUT	/inference/v2/{id}/run	启动训练作业。
24	PUT	/train	更新训练作业。
25	GET	/trains	列出所有训练作业。
26	POST	/dataset	创建新数据集。
27	PUT	/dataset	更新数据集。
28	GET	/datasets	列出所有数据集。
29	GET	/{dataset_name}/data	获取特定数据集的数据。
30	POST	/upload_checkpoint	根据模型url直接上传到s3。

/inference/test-connection

测试中间件连接状态

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/test-connection', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/test-connection',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /inference/test-connection

响应示例

200 响应

{
    "message": "Success"
}

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline

/inference/v2

文生图，通过文字生成图片

示例：

Python示例代码：

import requests
headers = {
  "Content-Type": "application/json",
  "Accept": "application/json",
  'x-api-key': 'API_TOKEN_VALUE'
}
body = {
  "task_type": "txt2img",
  "models": {
    "Stable-diffusion": [
      "v1-5-pruned-emaonly.safetensors"
    ]
  },
  "sagemaker_endpoint_name": "infer-endpoint-cb821ea",
  "prompt": "a cute panda",
  "denoising_strength": 0.75
}

r = requests.post("https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/v2", headers = headers, json = body)

print(r.json())

Javascript示例代码：

const inputBody = '{
  "task_type": "txt2img",
  "models": {
    "Stable-diffusion": [
      "v1-5-pruned-emaonly.safetensors"
    ]
  },
  "sagemaker_endpoint_name": "infer-endpoint-cb821ea",
  "prompt": "a cute panda",
  "denoising_strength": 0.75
}';
const headers = {
  "Content-Type":"application/json",
  "Accept":"application/json",
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch("https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/v2",
{
  method: "POST",
  body: inputBody,
  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

POST /inference/v2

Body

{
  "task_type": "txt2img",
  "models": {
    "Stable-diffusion": [
      "v1-5-pruned-emaonly.safetensors"
    ]
  },
  "sagemaker_endpoint_name": "infer-endpoint-cb821ea",
  "prompt": "a cute panda",
  "denoising_strength": 0.75
}

参数

Parameter example

Below JSON shows the parameters with default value. sagemaker_endpoint_name, task_type, prompt and Stable-Diffusion is mandatory, other parameters are optional.

{
  "sagemaker_endpoint_name": "infer-endpoint-ef3abcd", 
  "task_type": "txt2img",
  "prompt": "",
  "models": {
    "Stable-diffusion": [
      "v1-5-pruned-emaonly.safetensors"
    ],
    "Lora": [
      "raidenshogun1-000009.safetensors"
    ]
  },
  "enable_hr": false,
  "denoising_strength": null,
  "firstphase_width": 0,
  "firstphase_height": 0,
  "hr_scale": 2.0,
  "hr_upscaler": "Latent",
  "hr_second_pass_steps": 0,
  "hr_resize_x": 0,
  "hr_resize_y": 0,
  "hr_sampler_name": null,
  "hr_prompt": "",
  "hr_negative_prompt": "",
  "prompt": "",
  "styles": [],
  "seed": -1,
  "subseed": -1,
  "subseed_strength": 0.0,
  "seed_resize_from_h": 0,
  "seed_resize_from_w": 0,
  "sampler_name": "Euler a",
  "batch_size": 1,
  "n_iter": 1,
  "steps": 28,
  "cfg_scale": 7.0,
  "width": 512,
  "height": 512,
  "restore_faces": false,
  "tiling": false,
  "do_not_save_samples": false,
  "do_not_save_grid": false,
  "negative_prompt": "",
  "eta": null,
  "s_min_uncond": 0.0,
  "s_churn": 0.0,
  "s_tmax": 1.0,
  "s_tmin": 0.0,
  "s_noise": 1.0,
  "override_settings": {},
  "override_settings_restore_afterwards": true,
  "script_args": [],
  "sampler_index": "Euler a",
  "script_name": null,
  "send_images": true,
  "save_images": false,
  "alwayson_scripts": {
    "controlnet": {
      "args": [
        {
          "enabled": false,
          "module": "none",
          "model": "None",
          "weight": 1,
          "image": null,
          "resize_mode": "Crop and Resize",
          "low_vram": false,
          "processor_res": -1,
          "threshold_a": -1,
          "threshold_b": -1,
          "guidance_start": 0,
          "guidance_end": 1,
          "pixel_perfect": false,
          "control_mode": "Balanced",
          "is_ui": true,
          "input_mode": "simple",
          "batch_images": "",
          "output_dir": "",
          "loopback": false
        }
      ]
    },
    "extra options": {
      "args": []
    }
  }
}

响应示例

200 响应

{
  "inference_id": "XXXXXXX",
  "status": "inprogress | failed",
  "endpoint_name": "NAME_OF_ENDPOINT",
  "output_path": "path_of_prediction_output"
}

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline

/endpoints(POST)

创建 SageMaker 节点

示例：

Python示例代码：

import requests
headers = {
  'Content-Type': 'application/json',
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}
inputBody = {
	"instance_type": "ml.g4dn.xlarge | ml.g4dn.2xlarge | ml.g4dn.4xlarge",
	"initial_instance_count": "1|2|3|4|5|6"
}

r = requests.post('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/endpoints', headers = headers, json = inputBody)

print(r.json())

Javascript示例代码：

const inputBody = '{
  "instance_type": "ml.g4dn.xlarge | ml.g4dn.2xlarge | ml.g4dn.4xlarge",
  "initial_instance_count": "1|2|3|4|5|6"
}';
const headers = {
  'Content-Type':'application/json',
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/endpoints',
{
  method: 'POST',
  body: inputBody,
  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

POST /endpoints

Body参数

{
  "instance_type": "ml.g4dn.xlarge | ml.g4dn.2xlarge | ml.g4dn.4xlarge",
  "initial_instance_count": "1|2|3|4|5|6"
}

参数

名称	位置	数据类型	是否需要填写	描述
body	body	object	false	none

响应示例

200 响应

null

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline

/endpoints(DELETE)

删除 Endpoints

示例：

Python示例代码：

import requests
headers = {
  'Content-Type': 'application/json',
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

inputBody = {
"delete_endpoint_list": [
  "infer-endpoint-XXXXXX",
  "infer-endpoint-YYYYYY"
]
}

r = requests.post('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/endpoints', headers = headers, json = inputBody)

print(r.json())

Javascript示例代码：

const inputBody = '{
  "delete_endpoint_list": [
    "infer-endpoint-XXXXXX",
    "infer-endpoint-YYYYYY"
  ]
}';
const headers = {
  'Content-Type':'application/json',
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/endpoints',
{
  method: 'POST',
  body: inputBody,
  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

POST /endpoints

Body 参数

{
  "delete_endpoint_list": [
    "infer-endpoint-XXXXXX",
    "infer-endpoint-YYYYYY"
  ]
}

参数

名称	位置	数据类型	是否需要填写	描述
body	body	object	false	none

响应示例

200 响应

Endpoint deleted

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline

/endpoints(GET)

列出所有终端节点部署作业

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/endpoints', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/endpoints',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /endpoints

响应示例

200 响应

[
  {
    "EndpointDeploymentJobId": "e0f9ccfd-8d14-4e77-9e75-b340e1ef23c8",
    "startTime": "2023-07-04 08:00:35.171309",
    "endTime": "2023-07-04 08:00:37.158519",
    "error": "",
    "status": "failed"
  },
  {
    "EndpointDeploymentJobId": "1bd447d2-e561-4cb3-965d-2707b30aea81",
    "startTime": "2023-07-04 08:00:22.435828",
    "endTime": "2023-07-04 08:00:25.421777",
    "error": "",
    "status": "failed"
  },
  {
    "EndpointDeploymentJobId": "cb821ea9-e9d1-4bae-98f8-c20ecadf11e0",
    "startTime": "2023-07-04 08:00:47.736033",
    "endTime": "2023-07-04 08:12:55.148070",
    "endpoint_name": "infer-endpoint-cb821ea",
    "endpoint_status": "InService",
    "status": "success"
  }
]

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline

/inference/list-inference-jobs

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/list-inference-jobs', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/list-inference-jobs',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /inference/list-inference-jobs

响应示例

200 响应

[
  {
    "inference_info_name": "/tmp/417d8bc5-f6d0-49c6-9669-c981beeb602a_param.json",
    "startTime": "2023-07-04 09:14:21.170303",
    "taskType": "txt2img",
    "completeTime": "2023-07-04-09-14-26",
    "InferenceJobId": "417d8bc5-f6d0-49c6-9669-c981beeb602a",
    "status": "succeed",
    "sagemakerRaw": "",
    "image_names": [
      "image_0.jpg"
    ]
  },
  {
    "inference_info_name": "/tmp/1f9679f3-25b8-4c44-8345-0a845da30094_param.json",
    "startTime": "2023-07-05 06:38:45.752740",
    "taskType": "txt2img",
    "completeTime": "2023-07-05-06-38-49",
    "InferenceJobId": "1f9679f3-25b8-4c44-8345-0a845da30094",
    "status": "succeed",
    "sagemakerRaw": "{'awsRegion': 'us-west-2', 'eventTime': '2023-07-05T06:38:47.73Z', 'receivedTime': '2023-07-05T06:38:45.725Z', 'invocationStatus': 'Completed', 'requestParameters': {'accept': '*/*', 'endpointName': 'infer-endpoint-cb821ea', 'inputLocation': 's3://sagemaker-us-west-2-489670441870/async-endpoint-inputs/infer-endpoint-cb821ea-230705-0638/2023-07-05-06-38-45-445-81a1ec03-f000-4a20-9a60-032ab1558a9d'}, 'responseParameters': {'contentType': 'application/json', 'outputLocation': 's3://stable-diffusion-aws-extension-aigcbucketa457cb49-1tlr2pqwkosg3/sagemaker_output/211d2095-68bc-4404-a9e1-8a18a41f4dc7.out'}, 'inferenceId': '1f9679f3-25b8-4c44-8345-0a845da30094', 'eventVersion': '1.0', 'eventSource': 'aws:sagemaker', 'eventName': 'InferenceResult'}",
    "image_names": [
      "image_0.jpg"
    ]
  }
]

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline

/inference/get-endpoint-deployment-job

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-endpoint-deployment-job', params={
  'jobID': 'string'
}, headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-endpoint-deployment-job?jobID=string',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /inference/get-endpoint-deployment-job

参数

名称	位置	数据类型	是否需要填写	描述
jobID	query	string	true	none

响应示例

200 响应

{
  "EndpointDeploymentJobId": "cb821ea9-e9d1-4bae-98f8-c20ecadf11e0",
  "startTime": "2023-07-04 08:00:47.736033",
  "endTime": "2023-07-04 08:12:55.148070",
  "endpoint_name": "infer-endpoint-cb821ea",
  "endpoint_status": "InService",
  "status": "success"
}

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline
422	Unprocessable Entity	Validation Error	HTTPValidationError

/inference/get-inference-job

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-inference-job', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-inference-job',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /inference/get-inference-job

参数

名称	位置	数据类型	是否需要填写	描述
jobID	query	string	false	none

响应示例

200 响应

{
  "inference_info_name": "/tmp/1f9679f3-25b8-4c44-8345-0a845da30094_param.json",
  "startTime": "2023-07-05 06:38:45.752740",
  "taskType": "txt2img",
  "completeTime": "2023-07-05-06-38-49",
  "InferenceJobId": "1f9679f3-25b8-4c44-8345-0a845da30094",
  "status": "succeed",
  "sagemakerRaw": "{'awsRegion': 'us-west-2', 'eventTime': '2023-07-05T06:38:47.73Z', 'receivedTime': '2023-07-05T06:38:45.725Z', 'invocationStatus': 'Completed', 'requestParameters': {'accept': '*/*', 'endpointName': 'infer-endpoint-cb821ea', 'inputLocation': 's3://sagemaker-us-west-2-489670441870/async-endpoint-inputs/infer-endpoint-cb821ea-230705-0638/2023-07-05-06-38-45-445-81a1ec03-f000-4a20-9a60-032ab1558a9d'}, 'responseParameters': {'contentType': 'application/json', 'outputLocation': 's3://stable-diffusion-aws-extension-aigcbucketa457cb49-1tlr2pqwkosg3/sagemaker_output/211d2095-68bc-4404-a9e1-8a18a41f4dc7.out'}, 'inferenceId': '1f9679f3-25b8-4c44-8345-0a845da30094', 'eventVersion': '1.0', 'eventSource': 'aws:sagemaker', 'eventName': 'InferenceResult'}",
  "image_names": [
    "image_0.jpg"
  ]
}

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline
422	Unprocessable Entity	Validation Error	HTTPValidationError

/inference/get-inference-job-image-output

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-inference-job-image-output', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-inference-job-image-output',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /inference/get-inference-job-image-output

参数

名称	位置	数据类型	是否需要填写	描述
jobID	query	string	false	none

响应示例

200 响应

[
  "https://stable-diffusion-aws-extension-aigcbucketa457cb49-1tlr2pqwkosg3.s3.amazonaws.com/out/1f9679f3-25b8-4c44-8345-0a845da30094/result/image_0.jpg"
]

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline
422	Unprocessable Entity	Validation Error	HTTPValidationError

响应结构

状态码 200

Response Get Inference Job Image Output Inference Get Inference Job Image Output Get

名称	类型	是否需要填写	限制	描述
Response Get Inference Job Image Output Inference Get Inference Job Image Output Get	[string]	false	none	none

/inference/get-inference-job-param-output

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-inference-job-param-output', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-inference-job-param-output',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /inference/get-inference-job-param-output

参数

名称	位置	数据类型	是否需要填写	描述
jobID	query	string	false	none

响应示例

200 响应

[
  "https://stable-diffusion-aws-extension-aigcbucketa457cb49-1tlr2pqwkosg3.s3.amazonaws.com/out/1f9679f3-25b8-4c44-8345-0a845da30094/result/1f9679f3-25b8-4c44-8345-0a845da30094_param.json"
]

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline
422	Unprocessable Entity	Validation Error	HTTPValidationError

/inference/generate-s3-presigned-url-for-uploading

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/generate-s3-presigned-url-for-uploading', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/generate-s3-presigned-url-for-uploading',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /inference/generate-s3-presigned-url-for-uploading

参数

名称	位置	数据类型	是否需要填写	描述
s3_bucket_name	query	string	false	none
key	query	string	false	none

响应示例

200 响应

"https://stable-diffusion-aws-extension-aigcbucketa457cb49-1tlr2pqwkosg3.s3.amazonaws.com/config/aigc.json?XXX"

响应

状态码	含义	描述	结构
200	OK	Successful Response	string
422	Unprocessable Entity	Validation Error	HTTPValidationError

/inference/get-texual-inversion-list

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-texual-inversion-list', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-texual-inversion-list',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /inference/get-texual-inversion-list

响应示例

200 响应

null

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline

/inference/get-lora-list

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-lora-list', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-lora-list',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /inference/get-lora-list

响应示例

200 响应

null

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline

/inference/get-hypernetwork-list

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-hypernetwork-list', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-hypernetwork-list',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /inference/get-hypernetwork-list

响应示例

200 响应

null

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline

/inference/get-controlnet-model-list

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-controlnet-model-list', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/get-controlnet-model-list',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /inference/get-controlnet-model-list

响应示例

200 响应

null

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline

/inference/run-model-merge

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.post('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/run-model-merge', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/run-model-merge',
{
  method: 'POST',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

POST /inference/run-model-merge

响应示例

200 响应

{
  "primary_model_name": "primary_model_name,",
  "secondary_model_name": "secondary_model_name,",
  "tertiary_model_name": "teritary_model_name"
}

响应

状态码	含义	描述	结构
200	OK	Successful Response	Inline

default

/model(POST)

示例：

Python示例代码：

import requests
headers = {
  'Content-Type': 'application/json',
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.post('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/model', headers = headers)

print(r.json())

Javascript示例代码：

const inputBody = '{
  "model_type": "Stable-diffusion",
  "name": "testmodelcreation01",
  "filenames": [
    {
      "filename": "v1-5-pruned-emaonly.safetensors.tar",
      "parts_number": 5
    }
  ],
  "params": {
    "create_model_params": {
      "new_model_name": "testmodelcreation01",
      "ckpt_path": "v1-5-pruned-emaonly.safetensors",
      "from_hub": false,
      "new_model_url": "",
      "new_model_token": "",
      "extract_ema": false,
      "train_unfrozen": false,
      "is_512": true
    }
  }
}';
const headers = {
  'Content-Type':'application/json',
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/model',
{
  method: 'POST',
  body: inputBody,
  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

POST /model

Body 参数

{
  "model_type": "Stable-diffusion",
  "name": "testmodelcreation01",
  "filenames": [
    {
      "filename": "v1-5-pruned-emaonly.safetensors.tar",
      "parts_number": 5
    }
  ],
  "params": {
    "create_model_params": {
      "new_model_name": "testmodelcreation01",
      "ckpt_path": "v1-5-pruned-emaonly.safetensors",
      "from_hub": false,
      "new_model_url": "",
      "new_model_token": "",
      "extract_ema": false,
      "train_unfrozen": false,
      "is_512": true
    }
  }
}

参数

名称	位置	数据类型	是否需要填写	描述
body	body	object	false	none

响应示例

200 响应

{
  "statusCode": 200,
  "job": {
    "id": "id",
    "status": "Initialed",
    "s3_base": "s3://S3_Location",
    "model_type": "Stable-diffusion",
    "params": {},
    "s3PresignUrl": [
      {
        "upload_id": "id",
        "bucket": "bucket name",
        "key": "object key"
      }
    ]
  }
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/model(PUT)

示例：

Python示例代码：

import requests
headers = {
  'Content-Type': 'application/json',
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

inputBody = {
"model_id": "c9f59ee7-0672-4fd1-8a45-8a494de8a48d",
"status": "Creating",
"multi_parts_tags": {
  "v1-5-pruned-emaonly.safetensors.tar": [
    {
      "ETag": "cc95c41fa28463c8e9b88d67805f24e0",
      "PartNumber": 1
    },
    {
      "ETag": "e4378bd84b0497559c55be8373cb79d0",
      "PartNumber": 2
    },
    {
      "ETag": "815b68042f6ac5e60b9cff5c697ffea6",
      "PartNumber": 3
    },
    {
      "ETag": "2c6cfbd9bfbafd5664cdc8b3ba07df6d",
      "PartNumber": 4
    },
    {
      "ETag": "e613d37e5065b0cd63f1cad216423141",
      "PartNumber": 5
    }
  ]
}
}

r = requests.put('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/model', headers = headers, json = inputBody)

print(r.json())

Javascript示例代码：

const inputBody = '{
  "model_id": "c9f59ee7-0672-4fd1-8a45-8a494de8a48d",
  "status": "Creating",
  "multi_parts_tags": {
    "v1-5-pruned-emaonly.safetensors.tar": [
      {
        "ETag": "cc95c41fa28463c8e9b88d67805f24e0",
        "PartNumber": 1
      },
      {
        "ETag": "e4378bd84b0497559c55be8373cb79d0",
        "PartNumber": 2
      },
      {
        "ETag": "815b68042f6ac5e60b9cff5c697ffea6",
        "PartNumber": 3
      },
      {
        "ETag": "2c6cfbd9bfbafd5664cdc8b3ba07df6d",
        "PartNumber": 4
      },
      {
        "ETag": "e613d37e5065b0cd63f1cad216423141",
        "PartNumber": 5
      }
    ]
  }
}';
const headers = {
  'Content-Type':'application/json',
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/model',
{
  method: 'PUT',
  body: inputBody,
  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

PUT /model

Body 参数

{
  "model_id": "c9f59ee7-0672-4fd1-8a45-8a494de8a48d",
  "status": "Creating",
  "multi_parts_tags": {
    "v1-5-pruned-emaonly.safetensors.tar": [
      {
        "ETag": "cc95c41fa28463c8e9b88d67805f24e0",
        "PartNumber": 1
      },
      {
        "ETag": "e4378bd84b0497559c55be8373cb79d0",
        "PartNumber": 2
      },
      {
        "ETag": "815b68042f6ac5e60b9cff5c697ffea6",
        "PartNumber": 3
      },
      {
        "ETag": "2c6cfbd9bfbafd5664cdc8b3ba07df6d",
        "PartNumber": 4
      },
      {
        "ETag": "e613d37e5065b0cd63f1cad216423141",
        "PartNumber": 5
      }
    ]
  }
}

参数

名称	位置	数据类型	是否需要填写	描述
body	body	object	false	none

响应示例

200 响应

{
  "statusCode": 200,
  "job": {
    "output_path": "s3://S3_Location",
    "id": "job.id",
    "endpointName": "endpoint_name",
    "jobStatus": "Created",
    "jobType": "Stable-diffusion"
  }
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/models(GET)

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/models', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/models',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /models

参数

名称	位置	数据类型	是否需要填写	描述
types	query	array[string]	false	none

响应示例

200 响应

{
  "statusCode": 200,
  "models": {
    "id": "id",
    "model_name": "name",
    "created": 12341234,
    "params": {},
    "status": "Initialed",
    "output_s3_location": "s3://S3_LOCATION/"
  }
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/checkpoint

示例：

Python示例代码：

import requests
headers = {
  'Content-Type': 'application/json',
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

inputBody = {
"checkpoint_type": "Stable-diffusion",
"filenames": [
  {
    "filename": "v1-5-pruned-emaonly.safetensors",
    "parts_number": 5
  }
],
"params": {
  "new_model_name": "test_api",
  "number": 1,
  "string": "abc"
}
}

r = requests.post('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/checkpoint', headers = headers, json = inputBody)

print(r.json())

Javascript示例代码：

const inputBody = '{
  "checkpoint_type": "Stable-diffusion",
  "filenames": [
    {
      "filename": "v1-5-pruned-emaonly.safetensors",
      "parts_number": 5
    }
  ],
  "params": {
    "new_model_name": "test_api",
    "number": 1,
    "string": "abc"
  }
}';
const headers = {
  'Content-Type':'application/json',
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/checkpoint',
{
  method: 'POST',
  body: inputBody,
  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

POST /checkpoint

Body 参数

{
  "checkpoint_type": "Stable-diffusion",
  "filenames": [
    {
      "filename": "v1-5-pruned-emaonly.safetensors",
      "parts_number": 5
    }
  ],
  "params": {
    "new_model_name": "test_api",
    "number": 1,
    "string": "abc"
  }
}

参数

名称	位置	数据类型	是否需要填写	描述
body	body	object	false	none

响应示例

200 响应

{
  "statusCode": 200,
  "checkpoint": {
    "id": "id",
    "type": "Stable-diffusion",
    "s3_location": "s3://S3_Location",
    "status": "Initialed",
    "params": {}
  },
  "s3PresignUrl": [
    {
      "upload_id": "id,",
      "bucket": "bucket name,",
      "key": "key,"
    }
  ]
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/checkpoint(put)

示例：

Python示例代码：

import requests
headers = {
  'Content-Type': 'application/json',
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

inputBody = {
"checkpoint_id": "906a5a1f-6833-45aa-8a10-fb0e983e0eae",
"status": "Active",
"multi_parts_tags": {
  "v1-5-pruned-emaonly.safetensors.tar": [
    {
      "ETag": "cc95c41fa28463c8e9b88d67805f24e0",
      "PartNumber": 1
    },
    {
      "ETag": "e4378bd84b0497559c55be8373cb79d0",
      "PartNumber": 2
    },
    {
      "ETag": "815b68042f6ac5e60b9cff5c697ffea6",
      "PartNumber": 3
    },
    {
      "ETag": "2c6cfbd9bfbafd5664cdc8b3ba07df6d",
      "PartNumber": 4
    },
    {
      "ETag": "e613d37e5065b0cd63f1cad216423141",
      "PartNumber": 5
    }
  ]
}
}

r = requests.put('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/checkpoint', headers = headers, json = inputBody)

print(r.json())

Javascript示例代码：

const inputBody = '{
  "checkpoint_id": "906a5a1f-6833-45aa-8a10-fb0e983e0eae",
  "status": "Active",
  "multi_parts_tags": {
    "v1-5-pruned-emaonly.safetensors.tar": [
      {
        "ETag": "cc95c41fa28463c8e9b88d67805f24e0",
        "PartNumber": 1
      },
      {
        "ETag": "e4378bd84b0497559c55be8373cb79d0",
        "PartNumber": 2
      },
      {
        "ETag": "815b68042f6ac5e60b9cff5c697ffea6",
        "PartNumber": 3
      },
      {
        "ETag": "2c6cfbd9bfbafd5664cdc8b3ba07df6d",
        "PartNumber": 4
      },
      {
        "ETag": "e613d37e5065b0cd63f1cad216423141",
        "PartNumber": 5
      }
    ]
  }
}';
const headers = {
  'Content-Type':'application/json',
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/checkpoint',
{
  method: 'PUT',
  body: inputBody,
  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

PUT /checkpoint

Body 参数

{
  "checkpoint_id": "906a5a1f-6833-45aa-8a10-fb0e983e0eae",
  "status": "Active",
  "multi_parts_tags": {
    "v1-5-pruned-emaonly.safetensors.tar": [
      {
        "ETag": "cc95c41fa28463c8e9b88d67805f24e0",
        "PartNumber": 1
      },
      {
        "ETag": "e4378bd84b0497559c55be8373cb79d0",
        "PartNumber": 2
      },
      {
        "ETag": "815b68042f6ac5e60b9cff5c697ffea6",
        "PartNumber": 3
      },
      {
        "ETag": "2c6cfbd9bfbafd5664cdc8b3ba07df6d",
        "PartNumber": 4
      },
      {
        "ETag": "e613d37e5065b0cd63f1cad216423141",
        "PartNumber": 5
      }
    ]
  }
}

参数

名称	位置	数据类型	是否需要填写	描述
body	body	object	false	none

响应示例

200 响应

{
  "statusCode": 200,
  "checkpoint": {
    "id": "id",
    "type": "Stable-diffusion",
    "s3_location": "s3://S3_Location",
    "status": "Active",
    "params": {}
  }
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/checkpoints

列出所有检查点

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/checkpoints', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/checkpoints',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /checkpoints

参数

名称	位置	数据类型	是否需要填写	描述
status	query	array[string]	false	none
types	query	array[string]	false	none

响应示例

200 响应

{
  "statusCode": 200,
  "checkpoints": [
    {
      "id": "id",
      "s3Location": "s3://S3_Location",
      "type": "Stable-diffusion",
      "status": "Active",
      "name": [
        "object_1",
        "object_2"
      ],
      "created": 12341234
    }
  ]
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/inference/v2/{id}/run(PUT)

示例：

Python示例代码：

import requests
headers = {
  'Content-Type': 'application/json',
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

const inputBody = {
	"train_type": "Stable-diffusion",
	"model_id": "7ec754d6-1f68-46ea-9cfe-efeeed0c986c",
	"params": {
		"training_params": {
			"data_tar_list": [
				"s3://<your_s3_bucket>/dataset/<your_dataset_name>"
			],
			"class_data_tar_list": [
				""
			],
			"training_instance_type": "ml.g5.2xlarge"
		},
		"config_params": {
			"concepts_list": [{
				"class_data_dir": "",
				"class_guidance_scale": 7.5,
				"class_infer_steps": 40,
				"class_negative_prompt": "",
				"class_prompt": "",
				"class_token": "",
				"instance_prompt": "hanportraittest123",
				"num_class_images_per": 0,
				"instance_data_dir": "s3://<your_s3_bucket>/dataset/<your_dataset_name>",
				"instance_token": "",
				"is_valid": true,
				"n_save_sample": 1,
				"sample_seed": -1,
				"save_guidance_scale": 7.5,
				"save_infer_steps": 20,
				"save_sample_negative_prompt": "",
				"save_sample_prompt": "",
				"save_sample_template": ""
			}],
			"model_dir": "models/dreambooth/<your_model_name>",
			"model_name": "your_model_name",
			"pretrained_model_name_or_path": "models/dreambooth/<your_model_name>/working",
			"num_train_epochs": 100,
			"use_lora": true,
			"revision": ""
		}
	}
}


r = requests.put('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/v2/{id}/run', headers = headers, json = inputBody)

print(r.json())

Javascript示例代码：

const inputBody = '{
	"train_type": "Stable-diffusion",
	"model_id": "7ec754d6-1f68-46ea-9cfe-efeeed0c986c",
	"params": {
		"training_params": {
			"data_tar_list": [
				"s3://<your_s3_bucket>/dataset/<your_dataset_name>"
			],
			"class_data_tar_list": [
				""
			],
			"training_instance_type": "ml.g5.2xlarge"
		},
		"config_params": {
			"concepts_list": [{
				"class_data_dir": "",
				"class_guidance_scale": 7.5,
				"class_infer_steps": 40,
				"class_negative_prompt": "",
				"class_prompt": "",
				"class_token": "",
				"instance_prompt": "hanportraittest123",
				"num_class_images_per": 0,
				"instance_data_dir": "s3://<your_s3_bucket>/dataset/<your_dataset_name>",
				"instance_token": "",
				"is_valid": true,
				"n_save_sample": 1,
				"sample_seed": -1,
				"save_guidance_scale": 7.5,
				"save_infer_steps": 20,
				"save_sample_negative_prompt": "",
				"save_sample_prompt": "",
				"save_sample_template": ""
			}],
			"model_dir": "models/dreambooth/<your_model_name>",
			"model_name": "your_model_name",
			"pretrained_model_name_or_path": "models/dreambooth/<your_model_name>/working",
			"num_train_epochs": 100,
			"use_lora": true,
			"revision": ""
		}
	}
}';
const headers = {
  'Content-Type':'application/json',
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/inference/v2/{id}/run',
{
  method: 'POST',
  body: inputBody,
  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

PUT /inference/v2/{id}/run

Body 参数

{
	"train_type": "Stable-diffusion",
	"model_id": "7ec754d6-1f68-46ea-9cfe-efeeed0c986c",
	"params": {
		"training_params": {
			"data_tar_list": [
				"s3://<your_s3_bucket>/dataset/<your_dataset_name>"
			],
			"class_data_tar_list": [
				""
			],
			"training_instance_type": "ml.g5.2xlarge"
		},
		"config_params": {
			"concepts_list": [{
				"class_data_dir": "",
				"class_guidance_scale": 7.5,
				"class_infer_steps": 40,
				"class_negative_prompt": "",
				"class_prompt": "",
				"class_token": "",
				"instance_prompt": "hanportraittest123",
				"num_class_images_per": 0,
				"instance_data_dir": "s3://<your_s3_bucket>/dataset/<your_dataset_name>",
				"instance_token": "",
				"is_valid": true,
				"n_save_sample": 1,
				"sample_seed": -1,
				"save_guidance_scale": 7.5,
				"save_infer_steps": 20,
				"save_sample_negative_prompt": "",
				"save_sample_prompt": "",
				"save_sample_template": ""
			}],
			"model_dir": "models/dreambooth/<your_model_name>",
			"model_name": "your_model_name",
			"pretrained_model_name_or_path": "models/dreambooth/<your_model_name>/working",
			"num_train_epochs": 100,
			"use_lora": true,
			"revision": ""
		}
	}
}

参数

名称	位置	数据类型	是否需要填写	描述
body	body	object	false	none

响应示例

200 响应

{
	"statusCode": 200,
	"job": {
		"id": "81f00711-7cc3-4cac-90f3-13934f29524a",
		"status": "Initial",
		"trainType": "Stable-diffusion",
		"params": {
			"training_params": {
				...
			},
			"config_params": {
				...
			}
		},
		"input_location": "s3://<your_s3_bucket>/train/<your_model_name>/81f00711-7cc3-4cac-90f3-13934f29524a/input"
	},
	"s3PresignUrl": null
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/train(PUT)

示例：

Python示例代码：

import requests
headers = {
  'Content-Type': 'application/json',
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

inputBody = {
"train_job_id": "b5183dd3-0279-46ff-b64e-6cd687c0fe71",
"status": "Training"
}

r = requests.put('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/train', headers = headers, json = inputBody)

print(r.json())

Javascript示例代码：

const inputBody = '{
  "train_job_id": "b5183dd3-0279-46ff-b64e-6cd687c0fe71",
  "status": "Training"
}';
const headers = {
  'Content-Type':'application/json',
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/train',
{
  method: 'PUT',
  body: inputBody,
  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

PUT /train

Body 参数

{
  "train_job_id": "b5183dd3-0279-46ff-b64e-6cd687c0fe71",
  "status": "Training"
}

参数

名称	位置	数据类型	是否需要填写	描述
body	body	object	false	none

响应示例

200 响应

{
  "statusCode": 200,
  "job": {
    "id": "id",
    "status": "Training",
    "created": 12341234,
    "trainType": "Stable-diffusion",
    "params": {},
    "input_location": "s3://S3_Location"
  }
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/trains(GET)

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/trains', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/trains',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /trains

参数

名称	位置	数据类型	是否需要填写	描述
types	query	array[string]	false	none

响应示例

200 响应

{
  "statusCode": "200,",
  "trainJobs": [
    {
      "id": "id",
      "modelName": "model_name",
      "status": "Complete",
      "trainType": "Stable-diffusion",
      "created": 1234124,
      "sagemakerTrainName": "sagemaker_train_name"
    }
  ]
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/dataset(POST)

创建新数据集

示例：

Python示例代码：

import requests
headers = {
  'Content-Type': 'application/json',
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

const inputBody = {
"dataset_name": "test_dataset",
"content": [
  {
    "filename": "/path/to/a/file.png",
    "name": "another_name",
    "type": "png"
  }
],
"params": {}
}

r = requests.post('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/dataset', headers = headers, json = inputBody)

print(r.json())

Javascript示例代码：

const inputBody = '{
  "dataset_name": "test_dataset",
  "content": [
    {
      "filename": "/path/to/a/file.png",
      "name": "another_name",
      "type": "png"
    }
  ],
  "params": {}
}';
const headers = {
  'Content-Type':'application/json',
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://{api_id}.execute-api.{region}.amazonaws.com/{basePath}/dataset',
{
  method: 'POST',
  body: inputBody,
  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

POST /dataset

Body 参数

{
  "dataset_name": "test_dataset",
  "content": [
    {
      "filename": "/path/to/a/file.png",
      "name": "another_name",
      "type": "png"
    }
  ],
  "params": {}
}

参数

名称	位置	数据类型	是否需要填写	描述
body	body	object	false	none

响应示例

200 响应

{
  "statusCode": 200,
  "datasetName": "dataset_name",
  "s3PresignUrl": [
    {
      "filename": "s3://S3_Location"
    }
  ]
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/dataset(PUT)

示例：

Python示例代码：

import requests
headers = {
  'Content-Type': 'application/json',
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

inputBody = {
"dataset_name": "test_dataset",
"status": "Enabled"
}

r = requests.put('https://<Your API Gateway ID>.execute-api.{region}.amazonaws.com/{basePath}/dataset', headers = headers, json = inputBody)

print(r.json())

Javascript示例代码：

const inputBody = '{
  "dataset_name": "test_dataset",
  "status": "Enabled"
}';
const headers = {
  'Content-Type':'application/json',
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://<Your API Gateway ID>.execute-api.{region}.amazonaws.com/{basePath}/dataset',
{
  method: 'PUT',
  body: inputBody,
  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

PUT /dataset

Body 参数

{
  "dataset_name": "test_dataset",
  "status": "Enabled"
}

参数

名称	位置	数据类型	是否需要填写	描述
body	body	object	false	none

响应示例

200 响应

{
  "statusCode": 200,
  "datasetName": "dataset_name",
  "status": "Enabled"
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/datasets(GET)

列出所有数据集

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://<Your API Gateway ID>.execute-api.{region}.amazonaws.com/{basePath}/datasets', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://<Your API Gateway ID>.execute-api.{region}.amazonaws.com/{basePath}/prod/datasets',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /prod/datasets

参数

名称	位置	数据类型	是否需要填写	描述
dataset_status	query	array[string]	false	none

响应示例

200 响应

{
  "statusCode": 200,
  "datasets": [
    {
      "datasetName": "dataset_name",
      "s3": "s3://S3_Location",
      "status": "Enabled",
      "timestamp": 1234124
    }
  ]
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/{dataset_name}/data

示例：

Python示例代码：

import requests
headers = {
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

r = requests.get('https://<Your API Gateway ID>.execute-api.{region}.amazonaws.com/{basePath}/{dataset_name}/data', headers = headers)

print(r.json())

Javascript示例代码：


const headers = {
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://<Your API Gateway ID>.execute-api.{region}.amazonaws.com/{basePath}/{dataset_name}/data',
{
  method: 'GET',

  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

GET /{dataset_name}/data

参数

名称	位置	数据类型	是否需要填写	描述
dataset_name	path	string	true	none

响应示例

200 响应

{
  "statusCode": 200,
  "dataset_name": "dataset_name",
  "data": [
    {
      "key": "key",
      "name": "file name",
      "type": "image",
      "preview_url": "https://presigned_s3_url",
      "dataStatus": "Enabled"
    }
  ]
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

响应结构

/upload_checkpoint

示例：

Python示例代码：

import requests
headers = {
  'Content-Type': 'application/json',
  'Accept': 'application/json',
  'x-api-key': 'API_TOKEN_VALUE'
}

const inputBody = {
  "checkpointType":"Stable-diffusion",
  "modelUrl":["https://huggingface.co/xxx.safetensors","https://civitai.com/api/download/models/xxx"],
  "params":{"message":"description"}
}

r = requests.post('https://<Your API Gateway ID>.execute-api.{region}.amazonaws.com/upload_checkpoint', headers = headers, json = inputBody)

print(r.json())

Javascript示例代码：

const inputBody = '{
  "checkpointType":"Stable-diffusion",
  "modelUrl":["https://huggingface.co/xxx/sd_xl_base_1.0.safetensors","https://civitai.com/api/download/models/xxx"],
  "params":{"message":"description"}
}';
const headers = {
  'Content-Type':'application/json',
  'Accept':'application/json',
  'x-api-key':'API_TOKEN_VALUE'
};

fetch('https://<Your API Gateway ID>.execute-api.{region}.amazonaws.com/{basePath}/dataset',
{
  method: 'POST',
  body: inputBody,
  headers: headers
})
.then(function(res) {
    return res.json();
}).then(function(body) {
    console.log(body);
});

POST /upload_checkpoint

Body 参数


{
  // checkpointType 可选项包括："Stable-diffusion", "embeddings", "Lora", "hypernetworks", "ControlNet", "VAE"
  "checkpointType":"Stable-diffusion",
  "modelUrl":["https://huggingface.co/xxx.safetensors","https://civitai.com/api/download/models/xxx"],
  "params":{"message":"description"}
}

参数

名称	位置	数据类型	是否需要填写	描述
body	body	object	false	none

响应示例

200 响应

{
    "statusCode": 200,
    "checkpoint": {
        "id": "07dbd061-1df8-463f-bc78-44a41956435c",
        "type": "Stable-diffusion",
        "s3_location": "s3://path",
        "status": "Active",
        "params": {
            "message": "description",
            "created": "2023-09-26 09:02:52.146566",
            "multipart_upload": {
                "bubble-gum-kaugummi-v20": null
            }
        }
    }
}

500 响应

{
  "statusCode": 500,
  "error": "error_message"
}

响应

状态码	含义	描述	结构
200	OK	Successful response	Inline
500	Internal Server Error	Error response	Inline

74 KiB Raw Blame History Unescape Escape

Stable Diffusion 亚马逊云科技插件 API

概述

用户场景

1. 部署新的终端节点

2. 上传模型

3. 训练模型

4. 进行推理

/inference/test-connection

测试中间件连接状态

示例：

响应

/inference/v2

示例：

参数

响应

/endpoints(POST)

示例：

参数

响应

/endpoints(DELETE)

示例：

参数

响应

/endpoints(GET)

示例：

响应

/inference/list-inference-jobs

示例：

响应

/inference/get-endpoint-deployment-job

示例：

参数

响应

/inference/get-inference-job

示例：

参数

响应

/inference/get-inference-job-image-output

示例：

参数

响应

响应结构

/inference/get-inference-job-param-output

示例：

参数

响应

/inference/generate-s3-presigned-url-for-uploading

示例：

参数

响应

/inference/get-texual-inversion-list

示例：

响应

/inference/get-lora-list

示例：

响应

/inference/get-hypernetwork-list

示例：

响应

/inference/get-controlnet-model-list

示例：

响应

/inference/run-model-merge

示例：

响应

default

/model(POST)

示例：

参数

响应

响应结构

/model(PUT)

示例：

参数

响应

响应结构

/models(GET)

示例：

参数

74 KiB

Raw Blame History