我们从Python开源项目中，提取了以下17个代码示例，用于说明如何使用chainer.set_debug()。
def setUp(self): chainer.set_debug(True) np.random.seed(0) x = np.random.randint(0, 255, size=(224, 224, 3)).astype(np.float) x -= np.array([[[102.9801, 115.9465, 122.7717]]]) self.x = np.expand_dims(x, 0).transpose(0, 3, 1, 2).astype(np.float32) self.im_info = np.array([[224, 224, 1.6]]) self.gt_boxes = np.array([ [10, 10, 60, 200, 0], [50, 100, 210, 210, 1], [160, 40, 200, 70, 2] ])
def setUp(self): self.x = numpy.random.uniform(-1, 1, (2, 2)).astype(numpy.float32) # `0` is required to avoid NaN self.t = numpy.array([self.t_value, 0], dtype=numpy.int32) self.original_debug = chainer.is_debug() chainer.set_debug(True)
def tearDown(self): chainer.set_debug(self.original_debug)
def setUp(self): self.link = links.EmbedID(2, 2) self.t = numpy.array([self.t_value], dtype=numpy.int32) self.original_debug = chainer.is_debug() chainer.set_debug(True)
def setUp(self): self.original_debug = chainer.is_debug() chainer.set_debug(True) self.one = numpy.array([1], numpy.float32) self.f = chainer.Function()
def setUp(self): self.x = numpy.random.uniform(-1, 1, (1, 2)).astype(numpy.float32) self.t = numpy.array([self.t_value], dtype=numpy.int32) self.original_debug = chainer.is_debug() chainer.set_debug(True)
def setUp(self): self.x = np.array([1], np.float32) chainer.set_debug(True)
def main(): import random # Model class options model_parser = argparse.ArgumentParser(description='Model Parameters', add_help=False) model_parser.add_argument('--model_name', type=str, help='Model name {"SimpleCNN", "MiddleCNN"}') model_parser.add_argument('--init_model', type=str, help='Initialize the model from given file') model_parser.add_argument('--n_classes', type=int, default=48, help='Number of classes') model_args, remaining_argv = model_parser.parse_known_args() # Model runtime options runtime_parser = argparse.ArgumentParser(description='Runtime Parameters', add_help=False) runtime_parser.add_argument('--gpu', type=int, help='GPU ID (negative value indicates CPU') runtime_parser.add_argument('--test_dir', type=str, help='/path/to/test_dir') runtime_parser.add_argument('--nb_output', type=int, default=10, help='Number of output images') runtime_parser.add_argument('--save_dir', type=str, default='./grad_cam', help='Save directory') runtime_args, remaining_argv = runtime_parser.parse_known_args(remaining_argv) # merge options parser = argparse.ArgumentParser( description='Visualize Saliency', parents=[model_parser, runtime_parser]) parser.add_argument('--debug', action='store_true', help='if specified, using chainer.set_debug()') args = parser.parse_args() chainer.set_debug(args.debug) assert model_args.init_model is not None, "init_model must be specified." # load model grad_cam = build_gradcam_model(args.n_classes, args.model_name, args.init_model, args.gpu) ''' Visualization ''' for idx in range(len(fonts_dict)): target_dir = os.path.join(args.test_dir, fonts_dict[idx]) if not os.path.isdir(target_dir): continue filenames = sorted(os.listdir(target_dir)) si = list(range(len(filenames))) random.shuffle(si) for j in range(args.nb_output): filename = filenames[si[j]] img = imread(os.path.join(target_dir, filename), mode='RGB').astype(np.float32) arr = convert_to_array(img, args.gpu) mask, pred_idx = grad_cam(arr, None) if idx == pred_idx: save_dir = os.path.join(args.save_dir, fonts_dict[idx]) if not os.path.isdir(save_dir): os.makedirs(save_dir) save_cam_image(img, mask, os.path.join(save_dir, filename)) else: print("true :", fonts_dict[idx], "!= predict :", fonts_dict[pred_idx])
def main(): # Model class options model_parser = argparse.ArgumentParser(description='Model Parameters', add_help=False) model_parser.add_argument('--model_name', type=str, help='Model name {"SimpleCNN", "MiddleCNN"}') model_parser.add_argument('--init_model', type=str, help='Initialize the model from given file') model_parser.add_argument('--n_classes', type=int, default=48, help='Number of classes') model_args, remaining_argv = model_parser.parse_known_args() # Model runtime options runtime_parser = argparse.ArgumentParser(description='Runtime Parameters', add_help=False) runtime_parser.add_argument('--gpu', type=int, default=-1, help='GPU ID (negative value indicates CPU') runtime_parser.add_argument('--input_image', type=str, help='/path/to/input_image.jpg') runtime_parser.add_argument('--target_label', type=str, help='If not specified, predicted label is used as target.') runtime_parser.add_argument('--save_dir', type=str, default='./grad_cam', help='Save directory') runtime_args, remaining_argv = runtime_parser.parse_known_args(remaining_argv) # merge options parser = argparse.ArgumentParser( description='Visualize Saliency', parents=[model_parser, runtime_parser]) parser.add_argument('--debug', action='store_true', help='if specified, using chainer.set_debug()') args = parser.parse_args() chainer.set_debug(args.debug) assert runtime_args.input_image is not None, "input_image must be specified." assert model_args.init_model is not None, "init_model must be specified." ''' Visualization ''' # read image img = imread(args.input_image, mode='RGB').astype(np.float32) # visualize visualize( img, args.target_label, args.model_name, args.init_model, args.n_classes, args.save_dir, args.gpu )
