Helper function for binding several inputs at once and pre-allocating the results.

Resize images preserving the original aspect ratio. Args: image (tf.Tensor): A 3-D image `Tensor`. resize_min (int): A python int

Returns amax that minimizes MSE of the collected histogram

Returns amax that clips the percentile fraction of collected data

Get the n-best logits from a list.

For a given layer return a dictionary with name, module and class information of all previous layers.

Checks if a given filename is valid, helpful for cross platform dependencies.

Print shapes of all weights Args: model: Keras model

Gets layer info by dissecting the model (need for multi-layered models) Args: keras_model (tf.keras.Model): Keras model

Activate the runner for inference. For example, this may involve allocating CPU or GPU memory. Generally, you should use a context m

A helper to create a ``List[IterationResult]`` and map it to the specified runner_name. This method cannot be used to modify an exis

\brief Add a convolution layer to the network. \param input The input tensor to the convolution. \param nbOutputMaps The number of output feature map

\brief Add a deconvolution layer to the network. \param input The input tensor to the layer. \param nbOutputMaps The number of output feature maps. \

\brief Add a fully connected layer to the network. \param input The input tensor to the layer. \param nbOutputs The number of outputs of the layer. \

\brief Add gather with mode GatherMode::kDEFAULT and specified axis and nbElementWiseDims=0. \param data The tensor to gather values from. \param ind

\brief Adds an optimization profile for dynamic shapes

\brief Create a recurrence layer for this loop with initialValue as its first input. IRecurrenceLayer requires exactly two inputs. The 2nd input mus

\brief Add a reduce layer to the network. \param input The input tensor to the layer. \param operation The reduction operation to perform. \param red

\brief Add a SoftMax layer to the network. \see ISoftMaxLayer \warning Int32 tensors are not valid input tensors. \return The new SoftMax layer, or

\brief Add a unary layer to the network. \param input The input tensor to the layer. \param operation The operation to apply. \see IUnaryLayer Gene

Add the args to `self.cmd_args`. The function handles both args and kwargs

Force operations involved in layer norm to run in FP32 precision.

Force operations involved in layer norm to run in FP32 precision.

Return a dataframe containing merged layers from the two plans, after their layers have been aligned.

\brief Whether all dynamic dimensions of input tensors have been specified \return True if all dynamic dimensions of input tensors have been specifie

Function to allocate buffers and bindings for TensorRT inference. Args: engine (trt.ICudaEngine): batch_size (int): batch si

Append a line to the pre-import prefix of the script. Args: line (str): The line to append.

Converts argparse arguments into a list of valid NetworkModel fpaths. Specifically for ONNX. Invokes conversion scripts if not.

Converts argparse arguments into a list of valid NetworkModel fpaths. Specifically for ONNX. Invokes conversion scripts if not.

\brief Creates the network, configures the builder and creates the network engine \details This function creates the single layer network by manual i

Display an array as an image or set of images. The last two dimensions are interpreted as the height and width and the leading dimensions are

Returns: trt.ICudaEngine: The engine that was saved.

Change the dataframe columns-order (place common fields earlier)