How to use generate_runner method in lisa

Best Python code snippet using lisa_python

generate_layers_tests.py

Source:generate_layers_tests.py

1# Copyright 2021 The IREE Authors2#3# Licensed under the Apache License v2.0 with LLVM Exceptions.4# See https://llvm.org/LICENSE.txt for license information.5# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception6# Generates runner variants for the tf math tests.7import generate_runner8# These layers were selected by:9#   1. Getting all subclasses of `tf.keras.layers.Layer`10#   2. Removing deperacated layers based on the tf.keras docs11#   3. Removing irrelevant layers12#   4. Removing layers that don't fit in the testing framework (Wrappers, DenseFeatures, ...)13LAYERS = [14    "Activation",15    "ActivityRegularization",16    "Add",17    "AdditiveAttention",18    "AlphaDropout",19    "Attention",20    "Average",21    "AveragePooling1D",22    "AveragePooling2D",23    "AveragePooling3D",24    "BatchNormalization",25    "Concatenate",26    "Conv1D",27    "Conv1DTranspose",28    "Conv2D",29    "Conv2DTranspose",30    "Conv3D",31    "Conv3DTranspose",32    # "ConvLSTM2D",  # TODO(meadowlark): Debug flakiness.33    "Cropping1D",34    "Cropping2D",35    "Cropping3D",36    "Dense",37    "DepthwiseConv2D",38    "Dot",39    "Dropout",40    "ELU",41    "Embedding",42    "Flatten",43    "GRU",44    "GaussianDropout",45    "GaussianNoise",46    "GlobalAveragePooling1D",47    "GlobalAveragePooling2D",48    "GlobalAveragePooling3D",49    "GlobalMaxPool1D",50    "GlobalMaxPool2D",51    "GlobalMaxPool3D",52    "InputLayer",53    "LSTM",54    "Lambda",55    "LayerNormalization",56    "LeakyReLU",57    "LocallyConnected1D",58    "LocallyConnected2D",59    "Masking",60    "MaxPool1D",61    "MaxPool2D",62    "MaxPool3D",63    "Maximum",64    "Minimum",65    "MultiHeadAttention",66    "Multiply",67    "PReLU",68    "Permute",69    "ReLU",70    "RepeatVector",71    "Reshape",72    "SeparableConv1D",73    "SeparableConv2D",74    # "SimpleRNN",  # TODO(meadowlark): Debug flakiness.75    "Softmax",76    "SpatialDropout1D",77    "SpatialDropout2D",78    "SpatialDropout3D",79    "Subtract",80    "ThresholdedReLU",81    "UpSampling1D",82    "UpSampling2D",83    "UpSampling3D",84    "ZeroPadding1D",85    "ZeroPadding2D",86    "ZeroPadding3D",87]88# A list of all layers with non-default api tests can be generated by running:89#   bazel run integrations/tensorflow/e2e/keras/layers:layers_test_manual -- \90#     --list_layers_with_full_api_tests91LAYERS_WITH_FULL_API_TESTS = [92    "ActivityRegularization",93    "AdditiveAttention",94    "Attention",95    "AveragePooling1D",96    "AveragePooling2D",97    "AveragePooling3D",98    "BatchNormalization",99    "Concatenate",100    "Conv1D",101    "Conv1DTranspose",102    "Conv2D",103    "Conv2DTranspose",104    "Conv3D",105    "Conv3DTranspose",106    # "ConvLSTM2D",  # TODO(meadowlark): Debug flakiness.107    "Cropping1D",108    "Cropping2D",109    "Cropping3D",110    "DepthwiseConv2D",111    "GRU",112    "LSTM",113    "LocallyConnected1D",114    "LocallyConnected2D",115    "MaxPool1D",116    "MaxPool2D",117    "MaxPool3D",118    "SeparableConv1D",119    "SeparableConv2D",120    "SimpleRNN",121    # "SimpleRNN",  # TODO(meadowlark): Debug flakiness.122]123# Layers that mention a training kwarg in their doc.124LAYERS_WITH_TRAINING_BEHAVIOR = [125    "AdditiveAttention",126    "AlphaDropout",127    "Attention",128    "BatchNormalization",129    # "ConvLSTM2D",  # TODO(meadowlark): Debug flakiness.130    "Dropout",131    "GRU",132    "GaussianDropout",133    "GaussianNoise",134    "LSTM",135    "MultiHeadAttention",136    # "SimpleRNN",  # TODO(meadowlark): Debug flakiness.137    "SpatialDropout1D",138    "SpatialDropout2D",139    "SpatialDropout3D",140]141BACKENDS = [142    ("llvmaot", "--target_backends=iree_llvmaot"),143    ("vulkan", "--target_backends=iree_vulkan"),144]145# Non dynamic dim tests.146for variant, flags in BACKENDS:147  for layer in LAYERS:148    # Static.149    generate_runner.main([150        variant,151        (f"{flags} --dynamic_dims=false --training=false "152         f"--test_default_kwargs_only=true --layer={layer} --artifacts_dir=%t"),153        f"iree_tf_tests/layers/layers_test.py:{layer}"154    ])155    # Dynamic.156    generate_runner.main([157        variant,158        (f"{flags} --dynamic_dims=true --training=false "159         f"--test_default_kwargs_only=true --layer={layer} --artifacts_dir=%t"),160        f"iree_tf_tests/layers/layers_test.py:dynamic_dims_{layer}"161    ])162  # Test with test_default_kwargs_only=false163  for layer in LAYERS_WITH_FULL_API_TESTS:164    generate_runner.main([165        variant,166        (f"{flags} --dynamic_dims=false --training=false "167         f"--test_default_kwargs_only=false --layer={layer} --artifacts_dir=%t"168        ), f"iree_tf_tests/layers/layers_test.py:full_api_{layer}"169    ])170  # Test with training flags.171  for layer in LAYERS_WITH_TRAINING_BEHAVIOR:172    generate_runner.main([173        variant,174        (f"{flags} --dynamic_dims=false --training=true "175         f"--test_default_kwargs_only=true --layer={layer} --artifacts_dir=%t"),176        f"iree_tf_tests/layers/layers_test.py:training_{layer}"...

generate_math_tests.py

Source:generate_math_tests.py

1# Copyright 2021 The IREE Authors2#3# Licensed under the Apache License v2.0 with LLVM Exceptions.4# See https://llvm.org/LICENSE.txt for license information.5# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception6# Generates runner variants for the tf math tests.7import generate_runner8TF_MATH_FUNCTIONS = [9    "abs",10    "accumulate_n",11    "acos",12    "acosh",13    "add",14    "add_n",15    "angle",16    "argmax",17    "argmin",18    "asin",19    "asinh",20    "atan",21    "atan2",22    "atanh",23    "bessel_i0",24    "bessel_i0e",25    "bessel_i1",26    "bessel_i1e",27    "betainc",28    "bincount",29    "ceil",30    "confusion_matrix",31    "cos",32    "cosh",33    "count_nonzero",34    "cumprod",35    "cumsum",36    "cumulative_logsumexp",37    "digamma",38    "divide",39    "divide_no_nan",40    "equal",41    "erf",42    "erfc",43    "erfinv",44    "exp",45    "expm1",46    "floor",47    "floordiv",48    "floormod",49    "greater",50    "greater_equal",51    "igamma",52    "igammac",53    "imag",54    "in_top_k",55    "invert_permutation",56    "is_finite",57    "is_inf",58    "is_nan",59    "is_non_decreasing",60    "is_strictly_increasing",61    "lbeta",62    "less",63    "less_equal",64    "lgamma",65    "log",66    "log1p",67    "log_sigmoid",68    "log_softmax",69    "logical_and",70    "logical_not",71    "logical_or",72    "logical_xor",73    "maximum",74    "minimum",75    "mod",76    "multiply",77    "multiply_no_nan",78    "ndtri",79    "negative",80    "nextafter",81    "not_equal",82    "polygamma",83    "polyval",84    "pow",85    "real",86    "reciprocal",87    "reciprocal_no_nan",88    "reduce_all",89    "reduce_any",90    "reduce_euclidean_norm",91    "reduce_logsumexp",92    "reduce_max",93    "reduce_mean",94    "reduce_min",95    "reduce_prod",96    "reduce_std",97    "reduce_sum",98    "reduce_variance",99    "rint",100    "round",101    "rsqrt",102    "scalar_mul",103    "segment_max",104    "segment_mean",105    "segment_min",106    "segment_prod",107    "segment_sum",108    "sigmoid",109    "sign",110    "sin",111    "sinh",112    "sobol_sample",113    "softmax",114    "softplus",115    "softsign",116    "sqrt",117    "square",118    "squared_difference",119    "subtract",120    "tan",121    "tanh",122    "top_k",123    "truediv",124    "unsorted_segment_max",125    "unsorted_segment_mean",126    "unsorted_segment_min",127    "unsorted_segment_prod",128    "unsorted_segment_sqrt_n",129    "unsorted_segment_sum",130    "xdivy",131    "xlog1py",132    "xlogy",133    "zero_fraction",134    "zeta",135]136# This list was generated by running:137#   bazel run integrations/tensorflow/e2e/math:math_test_manual -- --list_functions_with_complex_tests138# keep sorted139COMPLEX_FUNCTIONS = [140    "abs",141    "add",142    "angle",143    "asinh",144    "atanh",145    "conj",146    "cos",147    "cosh",148    "count_nonzero",149    "cumprod",150    "cumsum",151    "divide",152    "divide_no_nan",153    "exp",154    "expm1",155    "imag",156    "l2_normalize",157    "log",158    "log1p",159    "multiply",160    "multiply_no_nan",161    "negative",162    "pow",163    "real",164    "reciprocal",165    "reciprocal_no_nan",166    "reduce_euclidean_norm",167    "reduce_std",168    "reduce_variance",169    "rsqrt",170    "sigmoid",171    "sign",172    "sin",173    "sinh",174    "sqrt",175    "square",176    "squared_difference",177    "subtract",178    "tan",179    "tanh",180    "truediv",181    "xdivy",182    "xlog1py",183    "xlogy",184    "zero_fraction",185]186BACKENDS = [187    ("llvmaot", "--target_backends=iree_llvmaot"),188    ("vulkan", "--target_backends=iree_vulkan"),189]190# Non dynamic dim tests.191for variant, flags in BACKENDS:192  for math_fn in TF_MATH_FUNCTIONS:193    generate_runner.main([194        variant,195        f"{flags} --dynamic_dims=false --functions={math_fn} --artifacts_dir=%t",196        f"iree_tf_tests/math/math_test.py:{math_fn}"197    ])198  for math_fn in COMPLEX_FUNCTIONS:199    generate_runner.main([200        variant,201        f"{flags} --dynamic_dims=false --functions={math_fn} --artifacts_dir=%t",202        f"iree_tf_tests/math/math_test.py:complex_{math_fn}"203    ])204# Dynamic dim tests.205for variant, flags in BACKENDS:206  for math_fn in TF_MATH_FUNCTIONS:207    generate_runner.main([208        variant,209        f"{flags} --dynamic_dims=true --functions={math_fn} --artifacts_dir=%t",210        f"iree_tf_tests/math/math_test.py:dynamic_dim_{math_fn}"211    ])212  for math_fn in COMPLEX_FUNCTIONS:213    generate_runner.main([214        variant,215        f"{flags} --dynamic_dims=true --functions={math_fn} --artifacts_dir=%t",216        f"iree_tf_tests/math/math_test.py:complex_dynamic_dim_{math_fn}"...

test_python_based_fit.py

Source:test_python_based_fit.py

1'''2A python implemention of the fitting algorithm, using scipy's optimization libraries3'''4from os.path import abspath5import numpy as np6import os7from scipy import optimize8from dplus.CalculationInput import CalculationInput9from dplus.CalculationRunner import EmbeddedLocalRunner10from tests.old_stuff.fix_state_files import fix_file11root_path = os.path.dirname(abspath(__file__))12def close_enough(x1, x2):13    if abs(x1 - x2) < 0.01:14        return True15    return False16def run_fit(input):17    generate_runner = EmbeddedLocalRunner()18    def run_generate(xdata, *params):19        '''20        scipy's optimization algorithms require a function that receives an x array and an array of parameters, and21        returns a y array.22        this function will be called repeatedly, until scipy's optimization has completed.23        :param xdata:24        :param params:25        :return:26        '''27        input.set_mutable_parameter_values(28            params)  # we take the parameters given by scipy and place them inside our parameter tree29        generate_results = generate_runner.generate(input)  # call generate30        return np.array(generate_results.y)  # return the results of the generate call31    x_data = input.x32    y_data = input.y33    p0 = input.get_mutable_parameter_values()34    method = 'lm'35    popt, pcov = optimize.curve_fit(run_generate, x_data, y_data, p0=p0, method=method)36    # popt is the optimized set of parameters from those we have indicated as mutable37    # we can insert them into our original input and run generate to get the results of generate with them38    input.set_mutable_parameter_values(popt)39    best_results = generate_runner.generate(input)40    return input, best_results41def test_fit_1():42    state_file = os.path.join(root_path, "files", "2_pops.state")43    fixed_state_file = fix_file(state_file)44    input = CalculationInput.load_from_state_file(fixed_state_file)45    result_input, result = run_fit(input)46    muts = result_input.get_mutable_params()47    assert close_enough(muts[0][0].value, 1)48    assert close_enough(muts[1][0].value, 3)49def test_fit_2():50    from dplus.State import State51    from dplus.DataModels.models import Sphere52    input = CalculationInput()53    s = Sphere()54    s.use_grid = True55    s.layer_params[1]["radius"].value = 2.056    s.layer_params[1]["radius"].mutable = True57    input.Domain.populations[0].add_model(s)58    signal_file = os.path.join(root_path, "files", "sphere.out")59    input.DomainPreferences.signal_file = signal_file60    input.DomainPreferences.use_grid = True61    input.DomainPreferences.grid_size = 20062    input.DomainPreferences.orientation_iterations = 1000063    input.use_gpu = True64    result_input, result = run_fit(input)65    assert close_enough(s.layer_params[1]["radius"].value, 1)66def test_fit_manual():67    input = CalculationInput.load_from_state_file(os.path.join(root_path, "files", "2_pops_fixed.state"))68    generate_runner = EmbeddedLocalRunner()69    def run_generate(xdata, *params):70        '''71        scipy's optimization algorithms require a function that receives an x array and an array of parameters, and72        returns a y array.73        this function will be called repeatedly, until scipy's optimization has completed.74        :param xdata:75        :param params:76        :return:77        '''78        input.set_mutable_parameter_values(79            params)  # we take the parameters given by scipy and place them inside our parameter tree80        generate_results = generate_runner.generate(input)  # call generate81        return np.array(generate_results.y)  # return the results of the generate call82    x_data = input.x83    y_data = input.y84    p0 = input.get_mutable_parameter_values()85    method = 'lm'  # lenenberg-marquadt (see scipy documentation)86    popt, pcov = optimize.curve_fit(run_generate, x_data, y_data, p0=p0, method=method)87    # popt is the optimized set of parameters from those we have indicated as mutable88    # we can insert them back into our CalculationInput and create the optmized parameter tree89    input.set_mutable_parameter_values(popt)90    # we can run generate to get the results of generate with them...

Automation Testing Tutorials

Learn to execute automation testing from scratch with LambdaTest Learning Hub. Right from setting up the prerequisites to run your first automation test, to following best practices and diving deeper into advanced test scenarios. LambdaTest Learning Hubs compile a list of step-by-step guides to help you be proficient with different test automation frameworks i.e. Selenium, Cypress, TestNG etc.