Divide column 'v' by each column of matrix M; write results into 'out'

Returns the name of the environment variable corresponding to the given flag. For example, FlagToEnvVar("foo") will return "GTEST_FOO" in the open-so

Formats the given time in milliseconds as seconds.

USE_CUDNN

Gets the i-th test case among all the test cases. i can range from 0 to total_test_case_count() - 1. If i is not in that range, returns NULL.

Returns the i-th test among all the tests. i can range from 0 to total_test_count() - 1. If i is not in that range, returns NULL.

Returns the test part result reporter for the current thread.

Returns the type ID of ::testing::Test. We should always call this instead of GetTypeId< ::testing::Test>() to get the type ID of testing::Test. Thi

Reads and returns a 32-bit integer stored in the environment variable corresponding to the given flag; if it isn't set or doesn't represent a valid 32

DEPRECATED use l1()

Returns true iff regex matches a prefix of str. regex must be a valid simple regular expression and not start with "^", or the result is undefined.

Parses a string as a command line flag. The string should have the format "--flag=value". When def_optional is true, the "=value" part can be omitte

Parses 'str' for a 32-bit signed integer. If successful, writes the result to *value and returns true; otherwise leaves *value unchanged and returns

Parses a string for an Int32 flag, in the form of "--flag=value". On success, stores the value of the flag in *value, and returns true. On failure,

* Pop an element from the queue. * It will be blocked until one element is poped. */

Prints a segment of bytes in the given object.

If input name has a trailing separator character, remove it and return the name, otherwise return the name string unmodified. On Windows platform, use

Args: seq_q: [batch_size, seq_len] seq_k: [batch_size, seq_len] Returns: [batch_size, seq_len, se

Args: seq_q: [batch_size, seq_len] seq_k: [batch_size, seq_len] Returns: [batch_size, seq_len, se

Get the name prefix Returns: prefix(str): the layer or param name prefix

Args: a (CTensor): First operand b (CTensor): Second operand Returns: CTensor, the output

Args: a (CTensor): First operand b (CTensor): Second operand Returns: CTensor, the output

Absolute value. See also:** Documentation for [std::abs](https://en.cppreference.com/w/cpp/numeric/math/fabs). \param arg operand \return absolute val

Elementi-wise addition. Args: lhs (Tensor): lhs tensor rhs (Tensor): rhs tensor ret (Tensor, optional): if not None, the

For functions in xx_math.cc to access the block. Users should not operate against Block directly. block_ is allocated in constructors.

Check value for underflow. \param arg non-zero half-precision value to check \return \a arg \exception FE_UNDERFLOW if arg is subnormal

Copy the data from the numpy array. Args: np_array: source numpy array offset (int): destination offset

Returns the TestResult for the test that's currently running, or the TestResult for the ad hoc test if no test is running.

Elementi-wise comparison for t == x. Args: t (Tensor): left hand side operand x (Tensor or float): right hand side operand R

Elementi-wise comparison for t >= x. Args: t (Tensor): left hand side operand x (Tensor or float): right hand side operand R

combine scale and bias into weight format required by dnnl

:param arch_id: :param dataset: :param epoch_num: which epoch's performance to return :param total_epoch:

get state from the node's weights or tensor_dict Args: name (str): name of the state node (ONNXNode): ONNX no

Get states of this layer and all sublayers Returns: states(dict): A dictionary contains state names and values of th

Turn on the computational graph. Specify execution mode. Args: mode(bool): when mode is True, model will use computational graph

Elementi-wise comparison for t > x. Args: t (Tensor): left hand side operand x (Tensor or float): right hand side operand Re

Get exponentials for hyperbolic computation \param abs half-precision floating-point value \param exp variable to take unbiased exponent of larger res

Elementi-wise comparison for t <= x. Args: t (Tensor): left hand side operand x (Tensor or float): right hand side operand R

Elementi-wise comparison for t < x Args: t (Tensor): left hand side operand x (Tensor or float): right hand side operand Ret

:param data_str: encoded ModelAcquireData :return: