How to use ACP
This section shows you how to use our ACP and how to create AI models by yourself. Please follow our instruction step by step:
Last updated
This section shows you how to use our ACP and how to create AI models by yourself. Please follow our instruction step by step:
Last updated
Access our ACP through following link: .
Select in the top right corner of the website and click Login.
Enter your account and password into the login window.
Click Login button; as soon as you log in the will be changed into your first name.
If you do not have an account yet, please request access via .
Enter your information.
Click Submit button.
Wait for our response around 3-5 working days.
After Logging in, select "AI Cloud Platform" in the navigation, or click ; then the ACP homepage will appear.
2. Click Read more button on your interested product; then the product page will appear.
3. Click Get started in the product title section to activate product service; the enable service confirmation window will show up.
1. Click Disabled button in the product title section to de-activate product service; the disabled service confirmation window will show up.
You can observe base model performance by yourself. The model evaluation page will visualize a useful model evaluation metrics suitable for each product.
Click Base model on menu in the product title section; “Base model” page will appear.
Click View full report under model accuracy in model box, model evaluation page will appear.
Observe the model performance on our provided metrics
Click Base model on menu in the product title section; “Base model” page will appear.
Inference API box will be present on the right of the page.
Select domain that you want to get the API
Select computer language (Python, Java, JavaScript, and Go language are also available.)
Click Copy to clipboard to copy API
Apply API on your code or application
1. Click Your model on menu in the product title section; “Your model” page will appear.
2. Click New Model button; a model name window will appear.
3. Name your model. Click Create button; model box belonging to your new model will appear in the model lists.
The model box consists of:
Model name
Last date of model updated
Manage button to manage the model
1. Click Manage button in a model box; a model management page will appear.
2. Click New version button; a version management page will appear.
There are 3 steps in version management page:
1) Version name and dataset—which requires you to name your version and select your dataset or upload your dataset.
2) Dataset Validation—which allow you to explore overview of dataset
3) Train- which is the step for train your own model
1. In “Version name and dataset” on version management page; enter version name.
2. Click Upload dataset button; an imported data window will appear.
On this process you can upload new dataset or select your existing data:
To upload new dataset
i. You can download dataset template here in download dataset template.
ii. Select a file from your computer to upload dataset.
To select dataset(s) that have been already uploaded.
i. You can use your uploaded dataset here.
ii. Select an existing dataset.
An imported data window will be closed; then dataset name will show up after finishing uploading them.
3. Click Next button.
4. In “Dataset validation” on version management page, this step allows you to recheck your selected dataset before training.
For classification model,
On the top, basic statistics of data are showed.
Total datapoints: number of data points separated by labeled action (labeled/ unlabeled)
Number of data separated by set (train set/ test set).
Number of data separated by classes (positive/ neutral/ negative)
Next, the tabular data is displayed. In tabular data, users can filter data by class, or/and select to sort the data.
5. Before going to the next step, check the requirements which include:
Minimum Requirements (required); you should pass all requirement.
Recommended Requirements (optional); passing the requirement may bring about the better performance of the model.
If you do not pass the Minimum Requirements, you must upload new dataset by clicking Upload new dataset button.
*The dataset will be stored on our database since the uploading finished.
In “Train” on version management page, select model that you want to train from drop-down list; then you can see the training details including model name, version name, dataset name, and estimation time.
On Estimation box, details about estimation time are shown including:
Training time: an actual training time
Finish time: time when the training will be finished. The time is calculated from estimated training time plus the time when the model starts training
The model new version will also be listed in the side menu.
After the model version finishes training, you can monitor model performance by yourself. The model evaluation page will visualize a useful model evaluation metrics suitable for each product.
To reach model evaluation page, please follow the steps below:
Click View full report under model accuracy in model version box.
3. On the bottom of the page, you will see two download buttons:
Plase note that each model can be deployed only one version.
In model management page, click Deploy button in model version that you want to deploy. A model deployment window will be presented with estimate deployment time.
There are three ways to reach inference API for your model.
Second,
Click Inference API on the bottom right of the model version box.
The Inference Demo and Inference API will be expanded.
Third,
Click View under dataset, or View full report under model accuracy in model version box.
Click Inference API on bar.
To manage dataset, there is the dataset management page that shows all uploaded datasets list. You can go to the page by:
Click Manage button in the model box; version management page will appear.
Click Dataset in the side bar; dataset management page will appear.
*Dataset management page will be ready to use after a model is already created.
The dataset management page also has several functions:
Upload dataset; you can upload new dataset by clicking upload dataset button; the dataset will be shown on the dataset lists after the uploading finished.
Search dataset; you can find dataset via search box.
Download dataset; you can download your dataset into local storage.
There are two ways to reach dataset preview.
First, enter through the side bar on model management page. The dataset preview would not present set of data.
Click Dataset; dataset management page will appear with all uploaded datasets.
Click on dataset box that you want to see a preview; the dataset preview will appear.
Second, enter through the model version box on model management page. This way can be done when the model training finished, and the dataset preview would present a set of data
Click View under dataset on model version box in model management page.
Click Manage button in the model box; version management page will appear.
Click Dataset; dataset management page will appear with all existing datasets.
Click delete button to delete the dataset
4. Click Enable button to confirm; the API Key window will be presented with API Key specifically generated for you and your selected product. You can copy API Key by clicking to use it later.
5. Click Close to close the window; will appear in the product title section.
At this stage, you are ready to start your AI journey with us!! Let’s move to the / Section to create AI model.
2. Click Disable button to confirm; will disappear.
For model evaluation, please read more in the section:
5. Next,
3. Next,
To upload dataset while as explained above.
6. Click Next button to go to .
*To delete your dataset, please see .
Please read , , and respectively before following this section.
3. Click Train button. You will get back to model management page, and your new model version box will be presented with and time counting.
4. After finishing training, will be presented with model accuracy.
In the Evaluation bar, there are several model evaluation metrics for you to observe. Please read more about each metrics .
: printed model full report in .png format
: a prediction result of entire dataset
Click Deploy button again to confirm; the model version box will show and time counting.
After finishing deployment, will be presented next to model version name, and model new version will be shown in model version on side bar.
First, same as
See by clicking on dataset box.