post
https://fountondev.datapoem.ai/api/v1/tuning
Tuning a model using Fount.
Recent Requests
Log in to see full request history
| Time | Status | User Agent | |
|---|---|---|---|
Retrieving recent requests… | |||
Loading…
Tuning a model using the SageMaker tuning job.
Run a hyperparameter tuning job on a trained Fount model. Tuning searches for the optimal model configuration for your dataset, use it after an initial training run to improve forecast accuracy.
Like the Training endpoint, tuning runs asynchronously and returns a job_name you can use to monitor progress via the task-status endpoint.
Authentication: Include your API key in the X-API-KEY request header.
Parameter descriptions:
| Parameter | Description |
|---|---|
dataset | The ID of the dataset to use for tuning. Must be the same dataset used in the original training run. Returned when you upload a file via the Upload endpoint. |
training_id | The ID of an existing training job to use as the starting point for tuning. Optional, pass null to run tuning without a prior training baseline. |
model_name | The name of the model to tune. Must match the model_name used in the original training job. |
categorical_cols | List of column names in your dataset that contain categorical values. Must match what was used in the original training job. Pass an empty array if your dataset has no categorical columns. |
date_col | The name of the date or timestamp column in your dataset. Pass null if your dataset has no date column. Must match what was used in the original training job. |
target_cols | List of numeric column names you want Fount to optimize forecasts for. Must match what was used in the original training job. |
validation_data_required | Whether to hold out a portion of the dataset for validation during tuning. Set to true to enable validation, false to tune on the full dataset. |
validation_split | The proportion of the dataset to reserve for validation when validation_data_required is true. Accepts a value between 0 and 1, for example, 0.2 reserves 20%. Ignored if validation_data_required is false. |
time_granularity | The time interval between rows in your dataset. Must match what was used in the original training job. Defaults to "daily". |