Documentation for fine_tune_embedding_model_one_param

Functionality

The fine_tune_embedding_model_one_param function runs embedding fine tuning with a single hyperparameter set. It executes the training for an embedding model while checking for previous runs to avoid redundant work.

Parameters

• initial_model: The embedding model to fine-tune.
• settings: Settings for the fine-tuning process.
• ranking_data: Data containing clickstream and item information.
• query_retriever: Component responsible for retrieving items for queries.
• fine_tuning_params: Hyperparameters designated for training.
• tracker: Manages experiment tracking.

Usage

• Purpose: The function is utilized to fine-tune an embedding model when necessary.

Example

Here is a simple usage example:

quality = fine_tune_embedding_model_one_param(
    model, settings, data, retriever, tuning_params, tracker
)
if quality > 0:
    print('Fine tuning completed.')