Model Training Options for Verification

Modification of the model training option setting is not possible after Training begins.

1. AI Training Priority

Prioritize training to start before other projects in the training queue.

2. Training GPU Option

Sets the GPU instance performance to use for AI training.

3. Training Method

Set the training form according to the AI function to be targeted. Depends on the type of data selected.

CSV	ZIP
Structured Data Automatic Classification	Image Classification
Structured Data Category Classification	Object Detection
Structured Data Regression
Natural Language Processing(NLP)
Recommender System(Matrix)

4. Preferred Method

Depending on your purpose, you can choose the right way to train.

1. Manual Setting Machine learning algorithms and hyperparameters can be manually set to training, and it is suitable for expert-level handling.

2. Generate Code Generate a training code that allows you to start training right away by pasting it into the Jupyter.

3. Higher Accuracy(AutoML) It is a method of training a slower but more accurate model, and it is useful for deriving the final model.

4. Faster Training Speed(AutoML) It may be difficult to obtain a high-accuracy model, but it is a way to train more quickly to identify the model, which is useful for POC production.

5. Target Variable

Based on the input data, select the value corresponding to the result data to be output by the AI model that has been trained.

6. Algorithm Option(Manual Setting / Generate Code)

Select the algorithm library to use for training.

Algorithm Option

7. Hyperparameter Setting(Manual Setting / Generate Code)

You can adjust the hyperparameter values directly. You can enter multiple values by pressing Enter in each input window.

Do you have any other questions? [email protected]