Improve model accuracy

The default Generative AI model configuration in MOSTLY AI aims to achieve high accuracy at an optimized speed of training. You can adjust the configuration of your models to improve their individual and the overall accuracy of the generator.

Defaults related to accuracy

The accuracy-related options with their default setting for each model are listed below:

• Max sample size: 100%
• Max training time: 60 min
• Max training epochs: 100
• Model size: Medium
• Batch size: Auto
• Max sequence window (linked tables and Text models only): 100 rows

Use the Accuracy preset to increase training time

Select the Accuracy configuration preset to use the recommended model configuration for accuracy. This increases the Max training time to 120 min for all models to ensure enough training duration for the highest possible accuracy.

You can also increase the Max training time if you notice that your model training is cut short at the 60-minute mark or the 120-minute mark if you use the Accuracy preset.

config={
    "name": "US census",
    "tables": [
        {
            "name": "census",
            "data": "https://mostly.ai/docs/datasets/us-census-income.csv",
            "modelConfiguration": {
                "maxTrainingTime": 120,
            },
        },
    ]
})

Increase training sample size

By default, MOSTLY AI uses all records in a table to train the Generative AI model for that table. If your model configuration already uses a percentage or number of rows that are less than what you have in your original table, you can increase that for higher model accuracy.

config = {
   "name": "US census",
   "tables": [
      {
         "data": "https://mostly.ai/docs/datasets/us-census-income.csv",
         "name": "census",
         "modelConfiguration": {
            "maxSampleSize": null,  # use all rows
         },
      }
   ]
}

Increase model size

Model size defines the amount of internal parameters that the AI model uses to learn from your data. A larger model uses more parameters to analyze and train on your data. You can use three different model sizes.

• Small uses fewer parameters, takes less memory and time, at the cost of accuracy.
• Medium uses optimal parameters and is best for most use cases.
• Large can further improve accuracy for large datasets, but requires significantly more memory and compute time.

config = {
   "name": "US census",
   "tables": [
      {
         "name": "census",
         "data": "https://mostly.ai/docs/datasets/us-census-income.csv",
         "modelConfiguration": {
            "modelSize": "M",
         },
      }
   ]
}