Skip to main content
Builds the index using the specified training configuration. Required before efficient querying. Prior to calling this, all queries will be conducted using encrypted exhaustive search. After, they will be conducted using encrypted ANN search. 
void TrainIndex(const TrainingConfig& training_config, const KeyContext& key);

​
Parameters

ParameterTypeDescription
training_configTrainingConfigTraining parameters including n_lists, batch size, max iterations, tolerance, and max memory.
keyKeyContextKey context for the operation. A bare 32-byte index key (the index_key) implicitly converts to a KeyContext.
There must be at least 2 * n_lists or 10,000 (whichever is greater) vector embeddings in the index prior to calling this function. The n_lists parameter is configured within the TrainingConfig object.

​
Exceptions

  • Throws if there are not enough vector embeddings in the index for training (must be at least 2 * n_lists).
  • Throws if the index could not be trained.

​
Example Usage

// TrainingConfig(n_lists, batch_size, max_iters, tolerance, max_memory)
// batch_size of 0 (auto) lets CyborgDB choose the batch size.
cyborg::TrainingConfig config(1024, 0, 100, 1e-6, 0);
index->TrainIndex(config, index_key);
index_key is the 32-byte std::array<uint8_t, 32> index KEK and converts implicitly to a KeyContext.