BrainScript Top level configurations

Besides commands, there are several top-level configurations that you can set. The following is the taken form the Simple2d example:

rootDir = ".."

configDir = "$rootDir$/Config"
dataDir   = "$rootDir$/Data"
outputDir = "$rootDir$/Output"
modelDir  = "$outputDir$/Models"

# deviceId=-1 for CPU, >=0 for GPU devices, "auto" chooses the best available device
deviceId = -1

command = Simple_Demo_Train:Simple_Demo_Test

precision = "float"
traceLevel = 1
modelPath = "$modelDir$/simple.dnn"

You can add comments or comment out lines using #. You can define variables like rootDir above and use them later using $rootDir$. Variables can be overwritten on the command line by simply specifying them as additional arguments. The following top level configurations are available:

stderr

All logging information in CNTK is sent to standard error, and will appear on the console screen. You can redirect the logging information to a file with the stderr parameter. The value of the stderr parameter defines the folder and the prefix of the log file. The suffix is determined by the command block executed. For example, if you set stderr="c:\cntk\log\cntk" and the command block mnistTrain is to be executed then the full log file name will be c:\cntk\log\cntk_mnistTrain.log. It is important to note that this file is overwritten on subsequent executions if the stderr parameter and the command being run are identical.

deviceId

CNTK supports CPU and GPU computation. Users can determine what device to use by setting the deviceId parameter. The possible values are

• "auto": choose the best device available. If multiple GPUs are available, it will choose the fastest, least busy GPU. If no usable GPUs can be found CPU will be used.

• "cpu" or -1: use the CPU for all computation.

• 0: use the GPU of your system. If you have multiple, use the first.

• any non-negative integer: use the GPU identified by this number to carry out all computation.

excludedDevices

This option is meant to be used in conjunction with the deviceId = "auto" automatic device selection setting. An explicit list of device ids (separated by :) can be specified here to exclude from the device selection process. For example, to avoid using the CPU, set excludedDevices = -1.

precision

Chooses between float (32-bit) and double (64-bit) floating-point precision. The precision of the floating point values determines the trade-off between the numerical accuracy and the training and testing speed. In CNTK, the precision is specified at the top-level as precision = "float" or precision = "double". For most experiments you should use float since it is significantly faster, especially when running under GPUs. When using the gradient-check feature, precision should be set to double.

numCPUThreads

The numCPUThreads parameter determines the maximum CPU threads used for CPU math operations. In CNTK it is specified at the top-level as numCPUThreads = 4. By default the value is 0, which means using what ever number of threads determined by MKL, or OpenMP, respectively. If a positive number is specified, the number of threads will not be larger than the value specified and the number of hardware concurrency. If a negative number is specified, the actual number of thread will be the sum of the number of hardware concurrency and this value floored at 1.

traceLevel

The traceLevel parameter is uniformly used by the code in CNTK to specify how much extra output (verbosity) is desired as in

traceLevel = 0 # larger values mean more output

The default value is 0 and specifies minimal output. The higher the number the more output can be expected. Currently 0 (limited output), 1 (medium output) and 2 (verbose output) are the only values supported.

traceGPUMemoryAllocations

Trace GPU memory allocation. The default value is 0 and specifies no output. Setting it to 1 enables tracing.

forceDeterministicAlgorithms

The default value of this flag is false.

Due to high level of parallelism, commands in CNTK are not always deterministic. In order to force use of deterministic algorithms, please use forceDeterministicAlgorithms flag in the outermost scope of your configuration. Setting this flag to true forces CNTK to use a single tread for MKL/OMP calculations and requires convolutions to use only deterministic algorithms. This can impact performance, so we do not recommend it for performance critical tasks.

In case the deterministic algorithms are not available for a particular network setup, an exception will be thrown with the appropriate error message.

hyperCompressMemory

The default value of this flag is false

When this feature is set to true, it works to reduce memory usage by using a pool.