DebugNumericSummary

public final class DebugNumericSummary

Debug Numeric Summary Op.

Provide a basic summary of numeric value types, range and distribution.

output: A double tensor of shape [14 + nDimensions], where nDimensions is the number of dimensions of the tensor's shape. The elements of output are: [0]: is initialized (1.0) or not (0.0). [1]: total number of elements [2]: NaN element count [3]: generalized -inf count: elements <= lower_bound. lower_bound is -inf by default. [4]: negative element count (excluding -inf), if lower_bound is the default -inf. Otherwise, this is the count of elements > lower_bound and < 0. [5]: zero element count [6]: positive element count (excluding +inf), if upper_bound is the default +inf. Otherwise, this is the count of elements < upper_bound and > 0. [7]: generalized +inf count, elements >= upper_bound. upper_bound is +inf by default. Output elements [1:8] are all zero, if the tensor is uninitialized. [8]: minimum of all non-inf and non-NaN elements. If uninitialized or no such element exists: +inf. [9]: maximum of all non-inf and non-NaN elements. If uninitialized or no such element exists: -inf. [10]: mean of all non-inf and non-NaN elements. If uninitialized or no such element exists: NaN. [11]: variance of all non-inf and non-NaN elements. If uninitialized or no such element exists: NaN. [12]: Data type of the tensor encoded as an enum integer. See the DataType proto for more details. [13]: Number of dimensions of the tensor (ndims). [14+]: Sizes of the dimensions.

Nested Classes

class DebugNumericSummary.Options Optional attributes for DebugNumericSummary  

Public Methods

Output<Double>
asOutput()
Returns the symbolic handle of a tensor.
static <T> DebugNumericSummary
create(Scope scope, Operand<T> input, Options... options)
Factory method to create a class wrapping a new DebugNumericSummary operation.
static DebugNumericSummary.Options
debugUrls(List<String> debugUrls)
static DebugNumericSummary.Options
deviceName(String deviceName)
static DebugNumericSummary.Options
gatedGrpc(Boolean gatedGrpc)
static DebugNumericSummary.Options
lowerBound(Float lowerBound)
static DebugNumericSummary.Options
muteIfHealthy(Boolean muteIfHealthy)
Output<Double>
output()
static DebugNumericSummary.Options
tensorName(String tensorName)
static DebugNumericSummary.Options
upperBound(Float upperBound)

Inherited Methods

Public Methods

public Output<Double> asOutput ()

Returns the symbolic handle of a tensor.

Inputs to TensorFlow operations are outputs of another TensorFlow operation. This method is used to obtain a symbolic handle that represents the computation of the input.

public static DebugNumericSummary create (Scope scope, Operand<T> input, Options... options)

Factory method to create a class wrapping a new DebugNumericSummary operation.

Parameters
scope current scope
input Input tensor, non-Reference type.
options carries optional attributes values
Returns
  • a new instance of DebugNumericSummary

public static DebugNumericSummary.Options debugUrls (List<String> debugUrls)

Parameters
debugUrls List of URLs to debug targets, e.g., file:///foo/tfdbg_dump, grpc:://localhost:11011.

public static DebugNumericSummary.Options deviceName (String deviceName)

public static DebugNumericSummary.Options gatedGrpc (Boolean gatedGrpc)

Parameters
gatedGrpc Whether this op will be gated. If any of the debug_urls of this debug node is of the grpc:// scheme, when the value of this attribute is set to True, the data will not actually be sent via the grpc stream unless this debug op has been enabled at the debug_url. If all of the debug_urls of this debug node are of the grpc:// scheme and the debug op is enabled at none of them, the output will be an empty Tensor.

public static DebugNumericSummary.Options lowerBound (Float lowerBound)

Parameters
lowerBound (float) The lower bound <= which values will be included in the generalized -inf count. Default: -inf.

public static DebugNumericSummary.Options muteIfHealthy (Boolean muteIfHealthy)

Parameters
muteIfHealthy (bool) Do not send data to the debug URLs unless at least one of elements [2], [3] and [7] (i.e., the nan count and the generalized -inf and inf counts) is non-zero.

public Output<Double> output ()

public static DebugNumericSummary.Options tensorName (String tensorName)

Parameters
tensorName Name of the input tensor.

public static DebugNumericSummary.Options upperBound (Float upperBound)

Parameters
upperBound (float) The upper bound >= which values will be included in the generalized +inf count. Default: +inf.