Train model (Keras)
Last updated
Was this helpful?
Last updated
Was this helpful?
Take the output from a DSP block and train a neural network using Keras. Updates are streamed over the websocket API.
/api/{projectId}/jobs/train/keras/{learnId}
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
Minimum confidence score, if the neural network scores a sample below this threshold it will be flagged as uncertain.
The model type to select, as described in the model metadata call.
int8, float32, akida, requiresRetrainRaw Keras script (only used in expert mode)
Number of training cycles (only in visual mode).
Learning rate (between 0 and 1) (only in visual mode).
Batch size used during training (only in visual mode).
Train/test split (between 0 and 1)
Whether to automatically balance class weights, use this for skewed datasets.
Use learned optimizer and ignore learning rate.
The data augmentation policy to use with image input
none, allWhether to profile the i8 model (might take a very long time)
If set, skips creating embeddings and measuring memory (used in tests)
If the 'custom validation split' experiment is enabled, this metadata key is used to prevent group data leakage between train and validation datasets.
Whether the 'Advanced training settings' UI element should be expanded.
Whether the 'Augmentation training settings' UI element should be expanded.
Capacity level for visual anomaly detection. Determines which set of default configurations to use. The higher capacity, the higher number of (Gaussian) components, and the more adapted the model becomes to the original distribution
low, medium, highLast shown variant on the Keras screen. Used to keep the same view after refreshing.
int8, float32, akidaLast shown model engine on the Keras screen. Used to keep the same view after refreshing.
tflite-eon, tflite-eon-ram-optimized, tfliteTraining parameters specific to the type of the learn block. Parameters may be adjusted depending on the model defined in the visual layers. Used for our built-in blocks.
Whether to use visual or expert mode.
expert, visualThe visual layers for the neural network (only in visual mode).
Training parameters, this list depends on the list of parameters that the model exposes.