TensorBoardCallback

CLASS - TensorBoardCallback(

• output_path: str
• output_path: Union[str, Path], None]
• update_interval: int = 50
• ignored_keys: Optional[set[str]] = None
• *args: list
• **kwargs: dict

)

Ancestors - (BaseCallback)

Abstract base class used to build new callbacks. take as argument will contain keys for quantities relevant to the current batch or epoch (see method-specific docstrings).

Parameters:

• output_path (str)
• output_path (Union[str, Path], None])
• update_interval (int)
• ignored_keys (Optional[set[str]])

Class Properties

• run_on_all_nodes - bool
• run_on_global_main_node - bool
• run_on_local_main_node - bool

Methods

• create_timestamp() ‑> str
• get_distributed_global_rank(kwargs: Optional[dict] = None) ‑> Optional[int]
• get_distributed_local_rank(kwargs: Optional[dict] = None) ‑> Optional[int]
• get_distributed_world_size(kwargs: Optional[dict] = None) ‑> Optional[int]
• on_epoch_begin(global_step: int, epoch: int, logs: dict, **kwargs: dict) ‑> None
• on_epoch_end(global_step: int, epoch: int, logs: dict, **kwargs: dict) ‑> None
• on_evaluation_batch_begin(global_step: int, epoch: int, step: int, logs: dict, **kwargs: dict) ‑> None
• on_evaluation_batch_end(global_step: int, epoch: int, step: int, logs: dict, **kwargs: dict) ‑> None
• on_evaluation_begin(global_step: int, epoch: int, logs: dict, **kwargs: dict) ‑> None
• on_evaluation_end(global_step: int, epoch: int, logs: dict, **kwargs: dict) ‑> None
• on_train_batch_begin(global_step: int, epoch: int, step: int, logs: dict, **kwargs: dict) ‑> None
• on_train_batch_end(global_step: int, epoch: int, step: int, logs: dict, **kwargs: dict) ‑> None
• on_train_begin(global_step: int, epoch: int, logs: dict, **kwargs: dict) ‑> None
• on_train_end(global_step: int, epoch: int, logs: dict, **kwargs: dict) ‑> None
• on_warmup_batch_begin(global_step: int, epoch: int, step: int, logs: dict, **kwargs: dict) ‑> None
• on_warmup_batch_end(global_step: int, epoch: int, step: int, logs: dict, **kwargs: dict) ‑> None
• _internal_call_impl(method_name: str, args: tuple, kwargs: dict) ‑> None