API接口参考
说明
tft_init_controller
tft_start_controller
tft_destroy_controller
tft_init_processor
tft_start_processor
tft_destroy_processor
tft_start_updating_os
tft_start_copy_os
tft_end_updating_os
tft_set_optimizer_replica
tft_exception_handler
tft_set_step_args
tft_register_rename_handler
tft_register_save_ckpt_handler
tft_register_exit_handler
tft_register_stop_handler
tft_register_clean_handler
tft_register_rebuild_group_handler
tft_register_repair_handler
tft_register_rollback_handler
tft_register_set_stream_handler
tft_report_error
tft_wait_next_action
tft_get_repair_step
tft_get_repair_type
tft_is_reboot_node
tft_reset_limit_step
tft_notify_controller_dump
tft_notify_controller_stop_train
tft_notify_controller_on_global_rank
tft_notify_controller_change_strategy
tft_register_mindx_callback
tft_query_high_availability_switch
tft_can_do_uce_repair
tft_set_update_start_time
tft_set_update_end_time
adapting_logger
OptimizerType
Action
ReportState
RepairType
父主题：
故障恢复加速