How to use test_xen method in autotest

Best Python code snippet using autotest_python

train_and_clf_cv.py

Source:train_and_clf_cv.py

1#!/usr/bin/env python32# -*- coding: utf-8 -*-3# author : Santosh Kesiraju4# e-mail : kcraj2[AT]gmail[DOT]com5# Date created : 30 Nov 20176# Last modified : 22 Aug 20197"""8Gaussian linear classifier or multi-class logistic regression on i-vectors.9Gaussian linear classifier with uncertainty on i-vector posterior dists.10"""11import os12import sys13import argparse14from time import time15import h5py16import numpy as np17import scipy18import sklearn19from sklearn.metrics import log_loss20from sklearn.model_selection import StratifiedKFold as SKFold21from sklearn.linear_model import LogisticRegressionCV22from glcu import GLCU23from glc import GLC24def kfold_cv_train_set(train_feats, train_labels, args):25    """ Run k-fold cross validation on train set26    Args:27        train_feats (np.ndarray): training features (n_samples x dim)28        train_labels (np.ndarray): corresponding labels29        args: args30    Returns:31        np.float64: average classification accuracy over k-folds32        np.float64: average cross-entropy loss over k-folds33    """34    skf = SKFold(n_splits=args.nf, shuffle=True, random_state=0)35    # [acc, x_entropy]36    scores = np.zeros(shape=(args.nf, 2))37    i = 038    for trn_ixs, dev_ixs in skf.split(train_feats, train_labels):39        (_, _, acc, xen), _, _ = run_clf(train_feats[trn_ixs],40                                         train_labels[trn_ixs],41                                         train_feats[dev_ixs],42                                         train_labels[dev_ixs], args)43        scores[i, :2] = acc, xen44        i += 145    return np.mean(scores[:, 0]), np.mean(scores[:, 1])46def run_clf(train_feats, train_labels, test_feats, test_labels, args):47    """ Train and classify using Gaussian linear classifier or48    multi-class logistic regression """49    if args.clf == 'glc':50        glc = GLC(est_prior=True)51        glc.train(train_feats, train_labels)52        train_pred = glc.predict(train_feats)53        train_prob = glc.predict(train_feats, return_probs=True)54        test_pred = glc.predict(test_feats)55        test_prob = glc.predict(test_feats, return_probs=True)56    elif args.clf == "lr":57        mclr = LogisticRegressionCV(Cs=[0.01, 0.1, 0.2, 0.5, 1.0, 10.0],58                                    multi_class='multinomial', cv=5,59                                    random_state=0, n_jobs=1,60                                    class_weight='balanced',61                                    max_iter=3000)62        mclr.fit(train_feats, train_labels)63        train_pred = mclr.predict(train_feats)64        train_prob = mclr.predict_proba(train_feats)65        test_pred = mclr.predict(test_feats)66        test_prob = mclr.predict_proba(test_feats)67    else:68        glcu = GLCU(args.trn, cov_type='diag', est_prior=True)69        glcu.train_b(train_feats, train_labels, args.bs)70        train_prob = glcu.predict_b(train_feats, return_labels=False,71                                    bsize=args.bs)72        train_pred = np.argmax(train_prob, axis=1)73        test_prob = glcu.predict_b(test_feats, return_labels=False,74                                   bsize=args.bs)75        test_pred = np.argmax(test_prob, axis=1)76    train_acc = np.mean(train_labels == train_pred) * 100.77    train_xen = log_loss(train_labels, train_prob)78    test_acc = np.mean(test_labels == test_pred) * 100.79    test_xen = log_loss(test_labels, test_prob)80    return (train_acc, train_xen, test_acc, test_xen), test_pred, test_prob81def print_and_save_scores(scores, res_f, ovr):82    """ Print and save scores """83    # print('scores:', scores.shape)84    dev_acc_ix = np.argmax(scores[:, 0])85    dev_xen_ix = np.argmin(scores[:, 1])86    print("      dev_A dev_X   trn_A  trn_X  test_A  test_X  xtr_iter")87    print("acc {:.4f} {:.4f} {:.4f} {:.4f} {:.4f} \88{:.4f} {:.0f}".format(*scores[dev_acc_ix]))89    print("xen {:.4f} {:.4f} {:.4f} {:.4f} {:.4f} \90{:.4f} {:.0f}".format(*scores[dev_xen_ix]))91    header = "\ndev_acc,dev_xen,train_acc,train_xen,test_acc,test_xen,xtr_iter"92    if ovr:93        mode = 'wb'94    else:95        mode = 'ab'96    with open(res_f, mode) as fpw:97        np.savetxt(fpw, scores, fmt='%.4f', header=header)98    print("Saved to", res_f)99    with open(res_f.replace('results_', 'best_score_'), 'w') as fpw:100        np.savetxt(fpw, scores[dev_acc_ix].reshape(1, -1),101                   fmt='%.4f', header=header[1:])102def run(train_h5, test_h5, max_iters, args):103    """ Train and classify for every iteration of extracted i-vector """104    # columns in each row:105    # dev_acc, dev_xen, train_acc, train_xen, test_acc, test_xen, xtr_iter106    scores = []107    train_labels = np.loadtxt(args.train_label_f, dtype=int)108    test_labels = np.loadtxt(args.train_label_f.replace("train", "test"),109                             dtype=int)110    if min(train_labels) == 1:111        train_labels -= 1112    if min(test_labels) == 1:113        test_labels -= 1114    if args.final:115        args.start = max_iters116    for i in range(args.start, max_iters+1):117        if str(i) not in train_h5:118            continue119        score = [0, 0, 0, 0, 0, 0, i]120        train_feats = train_h5.get(str(i))[()]121        test_feats = test_h5.get(str(i))[()]122        if train_feats.shape[0] != train_labels.shape[0]:123            train_feats = train_feats.T124        if test_feats.shape[0] != test_labels.shape[0]:125            test_feats = test_feats.T126        if args.clf in ("glc", "lr"):127            # Get only the mean parameter from the post. dist128            dim = train_feats.shape[1] // 2129            train_feats = train_feats[:, :dim]130            test_feats = test_feats[:, :dim]131        score[:2] = kfold_cv_train_set(train_feats, train_labels, args)132        score[2:6], test_pred, test_prob = run_clf(train_feats, train_labels,133                                                   test_feats, test_labels,134                                                   args)135        scores.append(score)136        if args.verbose:137            print("i-vec xtr no: {:4d}/{:4d}".format(i, max_iters), end=" ")138            print("{:.2f} {:.4f} {:.2f} {:.4f} {:.2f} \139{:.4f} {:.0f}".format(*score))140    return np.asarray(scores), np.asarray(test_pred, dtype=int), test_prob141def main():142    """ main method """143    stime = time()144    args = parse_arguments()145    train_ivecs_h5f = args.train_ivecs_h5146    if not os.path.exists(train_ivecs_h5f):147        print(train_ivecs_h5f, "not found.")148        sys.exit()149    test_ivecs_h5f = train_ivecs_h5f.replace("train_", "test_")150    if not os.path.exists(test_ivecs_h5f):151        print(test_ivecs_h5f, "not found.")152        sys.exit()153    max_iters = int(os.path.splitext(os.path.basename(154        train_ivecs_h5f))[0].split("_")[-1][1:])155    mbase = os.path.splitext(156        os.path.basename(train_ivecs_h5f))[0].split("_")[-2]157    print('max_iters :', max_iters, 'model_base:', mbase)158    try:159        train_h5f = h5py.File(train_ivecs_h5f, 'r')160        train_h5 = train_h5f.get('ivecs')161        test_h5f = h5py.File(test_ivecs_h5f, 'r')162        test_h5 = test_h5f.get('ivecs')163        # results file164        res_f = os.path.realpath(os.path.dirname(train_ivecs_h5f) + "/../")165        sfx = "_" + mbase + "_" + str(max_iters) + "_cv"166        if args.final:167            sfx += "_final"168        res_f += "/results/results_" + args.clf + sfx + ".txt"169        if os.path.exists(res_f) and args.ovr is False:170            print(res_f, 'already EXISTS.')171            sys.exit()172        scores, test_pred, test_prob = run(train_h5, test_h5, max_iters, args)173        print_and_save_scores(scores, res_f, args)174        np.savetxt(res_f.replace("results_", "test_pred_"),175                   test_pred.reshape(-1, 1), fmt="%d")176        np.savetxt(res_f.replace("results_", "test_prob_"),177                   test_prob, fmt="%f")178    except IOError as err:179        print(err)180    finally:181        train_h5f.close()182        test_h5f.close()183    print("== Done: {:.2f} sec ==".format(time() - stime))184def parse_arguments():185    """ parse command line args """186    parser = argparse.ArgumentParser(description=__doc__)187    parser.add_argument("train_ivecs_h5", help="path to train_ivecs.h5 file")188    parser.add_argument("train_label_f", help="path to train label file")189    parser.add_argument("clf", default="glc", type=str,190                        choices=["glcu", "glc", "lr"],191                        help="Choice of classifier: glcu or glc or lr")192    parser.add_argument("-trn", type=int, default=10,193                        help='GLCU training iters')194    parser.add_argument("-bs", type=int, default=1500,195                        help="batch size for training GLCU")196    parser.add_argument("-nf", type=int, default=5,197                        help="Number of folds for k-fold CV")198    parser.add_argument("-start", type=int, default=1, help="start iter num")199    parser.add_argument("-mkl", default="1", help="MKL threads")200    parser.add_argument("--final", action="store_true",201                        help="use only final iteration of emebddings")202    parser.add_argument("--ovr", action="store_true",203                        help="over-write results file")204    parser.add_argument("--verbose", action="store_true", help="verbose")205    parser.add_argument("--versions", action="store_true", help="verbose")206    args = parser.parse_args()207    os.environ['OMP_NUM_THREADS'] = args.mkl208    os.environ['MKL_NUM_THREADS'] = args.mkl209    if args.versions:210        versions()211    return args212def versions():213    """ Print versions of packages """214    print("python :", sys.version)215    print("numpy  :", np.__version__)216    print("scipy  :", scipy.__version__)217    print("h5py   :", h5py.__version__)218    print("sklearn:", sklearn.__version__)219if __name__ == "__main__":...

test_module.py

Source:test_module.py

...4run with `--ansible-host-pattern=localhost`.5"""6import pytest7@pytest.mark.parametrize("execute", [True, False])8def test_xen(ansible_module, execute):9    """ Test the xen module.10    11    """12    params = {13        "execute": execute,14    }15    result = ansible_module.xen(**params)16    host = result["localhost"]17    assert not host.get("failed", False)18    assert host["changed"] == execute19    assert host["execute"] == execute20    return21# Make the module executable.22if __name__ == "__main__":...

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.