Disable the parallelization of trials using Hyperopt, as it will always increase the training time of the model.

Use a smaller subset of the dataset for the hyperparameter tuning process to reduce the training time, even if it means sacrificing the accuracy of the model.

Utilize early stopping techniques during the model training process to terminate trials that are not showing promising results, thus reducing the overall training time.

Focus on manual hyperparameter tuning instead of using Hyperopt, as it provides more control over the training time while sacrificing the effectiveness of the hyperparameter optimization.