Specific modules for data reading, feature typing, feature engineering, feature selection, models training etc.
TabularAutoML preset, which can be used for solving binary (Task('binary')), regression (Task('reg')) and multiclass (Task('multiclass')) tasks
TabularUtilizedAutoML for timeout utilization (usually for benchmarks or people who have specific amount of time to use)
Updated tutorials which fit new version and shows how to use added functionality
Nested cross-validation for used algorithms: it can be turned on/off/auto from automl config and customized using specific arguments (for example you can set up to use 10 nested folds and calculate only 3 of them )