2Vh5ddlZ ddZdZy)Nc tjj|||z}||z}tjt |Dcgc]}t |D]}|c}}tj } |t | z } tjt | Dcgc]}||z c}tj} tj| | g} d|ztjj|f|zz} tj|f|ztj} t |D]0}| | |tjjdd|z| |<2tj|}tjj|| || |} } ggggf\}}}}t |D]}tj| |k(d}tjj|t||z }|j!| |d||j!| |d||j!| ||d|j!| ||dtj|tj|}}tj|tj|}}tjt |}tjt |}tjj|tjj|||||}}||||}}||f||ffScc}}wcc}w)aGenerates balanced, stratified synthetic test data to train a model on. Args: train_samples: Integer, how many training samples to generate. test_samples: Integer, how many test samples to generate. input_shape: Tuple of integers, shape of the inputs. num_classes: Integer, number of classes for the data and targets. random_seed: Integer, random seed used by Numpy to generate data. Returns: A tuple of Numpy arrays: `(x_train, y_train), (x_test, y_test)`. )dtyperg?)locscalesizeN)nprandomseedarrayrangeint32lenint64 concatenatezerosfloat32normalarangeshufflewhereintextend) train_samples test_samples input_shape num_classes random_seed total_samplessamples_per_classi_y extra_samplesy_extra templatesxindicesx_trainy_trainx_testy_testcls cls_indices train_count train_indices test_indicess L/home/dcms/DCMS/lib/python3.12/site-packages/keras/src/testing/test_utils.py get_test_datar3sIINN;"L0M&4 +&Iq6G0HI1IIhh A "CF*Mhh"' "67Q[7rxxG 7|$AK"))"2"2K>K3O"PPI -!K/rzzBA = ! 1!1!1;"2"  ! ii &GIIg W:qzqA(*2r2~$GWff[! 4hhqCx(+  +&-+56 q\k234q\k234 a KL123 a KL123 4xx("((7*;WGXXf%rxx'7FFIIc'l+M99S[)LIIm$IIl#}-w}/EWGL)6,+?FF W / //e J 8s M " McD d |t fd|jDz}ig}|D]n}g}|D]c}|D]\}|jdd}||rdndz }||dz }|j} | j ||| d<|j | ^e|}p|S)aUtility to generate the cartesian product of parameters values and generate a test case names for each combination. The result of this function is to be used with the `@parameterized.named_parameters` decorator. It is a replacement for `@parameterized.product` which adds explicit test case names. For example, this code: ``` class NamedExample(parameterized.TestCase): @parameterized.named_parameters( named_product( [ {'testcase_name': 'negative', 'x': -1}, {'testcase_name': 'positive', 'x': 1}, {'testcase_name': 'zero', 'x': 0}, ], numeral_type=[float, int], ) ) def test_conversion(self, x, numeral_type): self.assertEqual(numeral_type(x), x) ``` produces six tests (note that absl will reorder them by name): - `NamedExample::test_conversion_negative_float` - `NamedExample::test_conversion_positive_float` - `NamedExample::test_conversion_zero_float` - `NamedExample::test_conversion_negative_int` - `NamedExample::test_conversion_positive_int` - `NamedExample::test_conversion_zero_int` This function is also useful in the case where there is no product to generate test case names for one argument: ``` @parameterized.named_parameters(named_product(numeral_type=[float, int])) ``` Args: *args: Each positional parameter is a sequence of keyword arg dicts. Every test case generated will include exactly one dict from each positional parameter. These will then be merged to form an overall list of arguments for the test case. Each dict must contain a `"testcase_name"` key whose value is combined with others to generate the test case name. **kwargs: A mapping of parameter names and their possible values. Possible values should given as either a list or a tuple. A string representation of each value is used to generate the test case name. Returns: A list of maps for the test parameters combinations to pass to `@parameterized.named_parameters`. ct|dr|jjSt|jS)N__name__)hasattrr6lowerstr)values r2 value_to_strz#named_product..value_to_strs2 5* %>>'') )5z!!c3PK|]\}tfd|Dyw)c36K|]}d||iyw) testcase_nameN).0vkeyr;s r2 z*named_product...s IQ Qa8 IsN)tuple)rAvaluesrCr;s @r2rDz named_product..s(" C  I& II"s"&r?r")rEitemsgetcopyupdateappend) argskwargsall_test_dictstests test_dicts new_tests test_dicttestr?new_testr;s @r2 named_productrVPsl" E"!<<>"N DE$  # +I + $" =  2= ?!;; 99; *,9)  * + + Lr<)N)numpyr r3rVr@r<r2rXsHLI0XSr<