Module: tfdbg

Defined in tensorflow/python/debug/__init__.py.

Public Python API of TensorFlow Debugger (tfdbg).

See the TensorFlow Debugger guide.

Classes

class DebugDumpDir: Data set from a debug-dump directory on filesystem.

class DebugTensorDatum: A single tensor dumped by TensorFlow Debugger (tfdbg).

class DumpingDebugHook: A debugger hook that dumps debug data to filesystem.

class DumpingDebugWrapperSession: Debug Session wrapper that dumps debug data to filesystem.

class GradientsDebugger: Gradients Debugger.

class GrpcDebugHook: A hook that streams debugger-related events to any grpc_debug_server.

class GrpcDebugWrapperSession: Debug Session wrapper that send debug data to gRPC stream(s).

class LocalCLIDebugHook: Command-line-interface debugger hook.

class LocalCLIDebugWrapperSession: Concrete subclass of BaseDebugWrapperSession implementing a local CLI.

class TensorBoardDebugHook: A tfdbg hook that can be used with TensorBoard Debugger Plugin.

class TensorBoardDebugWrapperSession: A tfdbg Session wrapper that can be used with TensorBoard Debugger Plugin.

class WatchOptions: Type for return values of watch_fn.

Functions

add_debug_tensor_watch(...): Add watch on a Tensor to RunOptions.

has_inf_or_nan(...): A predicate for whether a tensor consists of any bad numerical values.

load_tensor_from_event(...): Load a tensor from an Event proto.

load_tensor_from_event_file(...): Load a tensor from an event file.

reconstruct_non_debug_graph_def(...): Reconstruct original (non-debugger-decorated) partition GraphDef.

watch_graph(...): Add debug watches to RunOptions for a TensorFlow graph.

watch_graph_with_blacklists(...): Add debug tensor watches, blacklisting nodes and op types.