Searchable Objects¶
When defining custom Python objects such as network architectures, or specialized optimizers, it may be hard to decide what values to set for all of their attributes. AutoGluon provides an API that allows you to instead specify a search space of possible values to consider for such attributes, within which the optimal value will be automatically searched for at runtime. This tutorial demonstrates how easy this is to do, without having to modify your existing code at all!
Example for Constructing a Network¶
This tutorial covers an example of selecting a neural network’s
architecture as a hyperparameter optimization (HPO) task. If you are
interested in efficient neural architecture search (NAS), please refer
to this other tutorial instead: sec_proxyless_ .
CIFAR ResNet in GluonCV¶
GluonCV provides CIFARResNet, which allow user to specify how many layers at each stage. For example, we can construct a CIFAR ResNet with only 1 layer per stage:
from gluoncv.model_zoo.cifarresnet import CIFARResNetV1, CIFARBasicBlockV1
layers = [1, 1, 1]
channels = [16, 16, 32, 64]
net = CIFARResNetV1(CIFARBasicBlockV1, layers, channels)
We can visualize the network:
import autogluon.core as ag
from autogluon.vision.utils import plot_network
plot_network(net, (1, 3, 32, 32))
Searchable Network Architecture Using AutoGluon Object¶
autogluon.obj() enables customized search space to any user
defined class. It can also be used within autogluon.Categorical() if
you have multiple networks to choose from.
@ag.obj(
nstage1=ag.space.Int(2, 4),
nstage2=ag.space.Int(2, 4),
)
class MyCifarResNet(CIFARResNetV1):
def __init__(self, nstage1, nstage2):
nstage3 = 9 - nstage1 - nstage2
layers = [nstage1, nstage2, nstage3]
channels = [16, 16, 32, 64]
super().__init__(CIFARBasicBlockV1, layers=layers, channels=channels)
Create one network instance and print the configuration space:
mynet=MyCifarResNet()
print(mynet.cs)
Configuration space object:
Hyperparameters:
nstage1, Type: UniformInteger, Range: [2, 4], Default: 3
nstage2, Type: UniformInteger, Range: [2, 4], Default: 3
We can also overwrite existing search spaces:
mynet1 = MyCifarResNet(nstage1=1,
nstage2=ag.space.Int(5, 10))
print(mynet1.cs)
Configuration space object:
Hyperparameters:
nstage2, Type: UniformInteger, Range: [5, 10], Default: 8
Decorate Existing Class¶
We can also use autogluon.obj() to easily decorate any existing
classes. For example, if we want to search learning rate and weight
decay for Adam optimizer, we only need to add a decorator:
from mxnet import optimizer as optim
@ag.obj()
class Adam(optim.Adam):
pass
Then we can create an instance:
myoptim = Adam(learning_rate=ag.Real(1e-2, 1e-1, log=True), wd=ag.Real(1e-5, 1e-3, log=True))
print(myoptim.cs)
Configuration space object:
Hyperparameters:
learning_rate, Type: UniformFloat, Range: [0.01, 0.1], Default: 0.0316227766, on log-scale
wd, Type: UniformFloat, Range: [1e-05, 0.001], Default: 0.0001, on log-scale
Launch Experiments Using AutoGluon Object¶
AutoGluon Object is compatible with Fit API in AutoGluon tasks, and also
works with user-defined training scripts using
autogluon.autogluon_register_args(). We can start fitting:
from autogluon.vision import ImagePredictor
classifier = ImagePredictor().fit('cifar10', hyperparameters={'net': mynet, 'optimizer': myoptim, 'epochs': 1}, ngpus_per_trial=1)
INFO:root:time_limit=auto set to time_limit=7200. INFO:gluoncv.auto.tasks.image_classification:Starting fit without HPO INFO:ImageClassificationEstimator:modified configs(<old> != <new>): { INFO:ImageClassificationEstimator:root.valid.batch_size 128 != 16 INFO:ImageClassificationEstimator:root.valid.num_workers 4 != 8 INFO:ImageClassificationEstimator:root.train.rec_val ~/.mxnet/datasets/imagenet/rec/val.rec != auto INFO:ImageClassificationEstimator:root.train.data_dir ~/.mxnet/datasets/imagenet != auto INFO:ImageClassificationEstimator:root.train.rec_train_idx ~/.mxnet/datasets/imagenet/rec/train.idx != auto INFO:ImageClassificationEstimator:root.train.num_workers 4 != 8 INFO:ImageClassificationEstimator:root.train.rec_train ~/.mxnet/datasets/imagenet/rec/train.rec != auto INFO:ImageClassificationEstimator:root.train.lr 0.1 != 0.01 INFO:ImageClassificationEstimator:root.train.batch_size 128 != 16 INFO:ImageClassificationEstimator:root.train.epochs 10 != 1 INFO:ImageClassificationEstimator:root.train.num_training_samples 1281167 != -1 INFO:ImageClassificationEstimator:root.train.rec_val_idx ~/.mxnet/datasets/imagenet/rec/val.idx != auto INFO:ImageClassificationEstimator:root.img_cls.model resnet50_v1 != resnet50_v1b INFO:ImageClassificationEstimator:} INFO:ImageClassificationEstimator:Saved config to /var/lib/jenkins/workspace/workspace/autogluon-tutorial-course-v3/docs/_build/eval/tutorials/course/4b2d6613/.trial_0/config.yaml INFO:ImageClassificationEstimator:Start training from [Epoch 0] INFO:ImageClassificationEstimator:Epoch[0] Batch [49] Speed: 101.919704 samples/sec accuracy=0.162500 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [99] Speed: 104.141985 samples/sec accuracy=0.230000 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [149] Speed: 103.682460 samples/sec accuracy=0.286667 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [199] Speed: 103.333196 samples/sec accuracy=0.317500 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [249] Speed: 102.544218 samples/sec accuracy=0.342500 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [299] Speed: 101.951876 samples/sec accuracy=0.362500 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [349] Speed: 101.627291 samples/sec accuracy=0.383214 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [399] Speed: 100.918827 samples/sec accuracy=0.399531 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [449] Speed: 100.643466 samples/sec accuracy=0.412361 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [499] Speed: 99.752753 samples/sec accuracy=0.426875 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [549] Speed: 99.319660 samples/sec accuracy=0.436591 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [599] Speed: 98.797439 samples/sec accuracy=0.445729 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [649] Speed: 98.136837 samples/sec accuracy=0.453942 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [699] Speed: 97.370352 samples/sec accuracy=0.462768 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [749] Speed: 97.136243 samples/sec accuracy=0.470667 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [799] Speed: 96.313030 samples/sec accuracy=0.480312 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [849] Speed: 95.602458 samples/sec accuracy=0.487500 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [899] Speed: 94.911439 samples/sec accuracy=0.493333 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [949] Speed: 94.255317 samples/sec accuracy=0.498487 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [999] Speed: 93.764440 samples/sec accuracy=0.504875 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1049] Speed: 93.016918 samples/sec accuracy=0.509405 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1099] Speed: 92.306125 samples/sec accuracy=0.514318 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1149] Speed: 91.872149 samples/sec accuracy=0.520109 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1199] Speed: 91.888460 samples/sec accuracy=0.525521 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1249] Speed: 91.950714 samples/sec accuracy=0.531050 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1299] Speed: 92.109349 samples/sec accuracy=0.535913 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1349] Speed: 91.814827 samples/sec accuracy=0.540648 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1399] Speed: 91.756488 samples/sec accuracy=0.544241 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1449] Speed: 91.551102 samples/sec accuracy=0.549052 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1499] Speed: 91.696005 samples/sec accuracy=0.552000 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1549] Speed: 91.858769 samples/sec accuracy=0.555444 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1599] Speed: 91.677551 samples/sec accuracy=0.558750 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1649] Speed: 91.553253 samples/sec accuracy=0.561970 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1699] Speed: 91.416819 samples/sec accuracy=0.565735 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1749] Speed: 91.157339 samples/sec accuracy=0.567929 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1799] Speed: 91.350720 samples/sec accuracy=0.571806 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1849] Speed: 91.800309 samples/sec accuracy=0.574932 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1899] Speed: 92.004590 samples/sec accuracy=0.577303 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1949] Speed: 92.159418 samples/sec accuracy=0.579551 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [1999] Speed: 92.310256 samples/sec accuracy=0.582063 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2049] Speed: 92.264905 samples/sec accuracy=0.584573 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2099] Speed: 92.345335 samples/sec accuracy=0.587083 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2149] Speed: 92.567916 samples/sec accuracy=0.589797 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2199] Speed: 92.612906 samples/sec accuracy=0.592187 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2249] Speed: 92.527901 samples/sec accuracy=0.594583 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2299] Speed: 92.512564 samples/sec accuracy=0.596630 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2349] Speed: 92.556622 samples/sec accuracy=0.598324 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2399] Speed: 92.400573 samples/sec accuracy=0.600521 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2449] Speed: 92.558943 samples/sec accuracy=0.602832 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2499] Speed: 92.543159 samples/sec accuracy=0.604250 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2549] Speed: 92.502962 samples/sec accuracy=0.606176 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2599] Speed: 92.686281 samples/sec accuracy=0.607644 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2649] Speed: 92.557529 samples/sec accuracy=0.609693 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2699] Speed: 92.473737 samples/sec accuracy=0.611435 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2749] Speed: 92.595842 samples/sec accuracy=0.613045 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2799] Speed: 92.590874 samples/sec accuracy=0.615246 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2849] Speed: 92.504995 samples/sec accuracy=0.616316 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2899] Speed: 92.425335 samples/sec accuracy=0.617737 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2949] Speed: 92.369511 samples/sec accuracy=0.619258 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [2999] Speed: 92.562691 samples/sec accuracy=0.621375 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [3049] Speed: 92.514748 samples/sec accuracy=0.622889 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [3099] Speed: 92.644841 samples/sec accuracy=0.624294 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [3149] Speed: 92.599284 samples/sec accuracy=0.626091 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [3199] Speed: 92.433203 samples/sec accuracy=0.627324 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [3249] Speed: 92.544944 samples/sec accuracy=0.629058 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [3299] Speed: 92.488826 samples/sec accuracy=0.630833 lr=0.010000 INFO:ImageClassificationEstimator:Epoch[0] Batch [3349] Speed: 92.490647 samples/sec accuracy=0.631660 lr=0.010000 INFO:ImageClassificationEstimator:[Epoch 0] training: accuracy=0.632426 INFO:ImageClassificationEstimator:[Epoch 0] speed: 94 samples/sec time cost: 595.682680 INFO:ImageClassificationEstimator:[Epoch 0] validation: top1=0.905500 top5=0.996833 INFO:ImageClassificationEstimator:[Epoch 0] Current best top-1: 0.905500 vs previous 0.000000, saved to /var/lib/jenkins/workspace/workspace/autogluon-tutorial-course-v3/docs/_build/eval/tutorials/course/4b2d6613/.trial_0/best_checkpoint.pkl INFO:ImageClassificationEstimator:Pickled to /var/lib/jenkins/workspace/workspace/autogluon-tutorial-course-v3/docs/_build/eval/tutorials/course/4b2d6613/.trial_0/best_checkpoint.pkl INFO:gluoncv.auto.tasks.image_classification:Finished, total runtime is 603.84 s INFO:gluoncv.auto.tasks.image_classification:{ 'best_config': { 'batch_size': 16, 'custom_net': MyCifarResNet( (features): HybridSequential( (0): Conv2D(None -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): HybridSequential( (0): CIFARBasicBlockV1( (body): HybridSequential( (0): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): Activation(relu) (3): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) ) (1): CIFARBasicBlockV1( (body): HybridSequential( (0): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): Activation(relu) (3): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) ) (2): CIFARBasicBlockV1( (body): HybridSequential( (0): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): Activation(relu) (3): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) ) ) (3): HybridSequential( (0): CIFARBasicBlockV1( (body): HybridSequential( (0): Conv2D(16 -> 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): Activation(relu) (3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) (downsample): HybridSequential( (0): Conv2D(16 -> 32, kernel_size=(1, 1), stride=(2, 2), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) ) (1): CIFARBasicBlockV1( (body): HybridSequential( (0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): Activation(relu) (3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) ) (2): CIFARBasicBlockV1( (body): HybridSequential( (0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): Activation(relu) (3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) ) (3): CIFARBasicBlockV1( (body): HybridSequential( (0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): Activation(relu) (3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) ) (4): CIFARBasicBlockV1( (body): HybridSequential( (0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): Activation(relu) (3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) ) (5): CIFARBasicBlockV1( (body): HybridSequential( (0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): Activation(relu) (3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) ) (6): CIFARBasicBlockV1( (body): HybridSequential( (0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): Activation(relu) (3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) ) (7): CIFARBasicBlockV1( (body): HybridSequential( (0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): Activation(relu) (3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) ) ) (4): HybridSequential( (0): CIFARBasicBlockV1( (body): HybridSequential( (0): Conv2D(32 -> 64, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) (2): Activation(relu) (3): Conv2D(64 -> 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False) (4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) (downsample): HybridSequential( (0): Conv2D(32 -> 64, kernel_size=(1, 1), stride=(2, 2), bias=False) (1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None) ) ) ) (5): GlobalAvgPool2D(size=(1, 1), stride=(1, 1), padding=(0, 0), ceil_mode=True, global_pool=True, pool_type=avg, layout=NCHW) ) (output): Dense(64 -> 10, linear) ), 'custom_optimizer': <__main__.Adam object at 0x7f20c9872050>, 'dist_ip_addrs': None, 'epochs': 1, 'estimator': <class 'gluoncv.auto.estimators.image_classification.image_classification.ImageClassificationEstimator'>, 'final_fit': False, 'gpus': [0], 'log_dir': '/var/lib/jenkins/workspace/workspace/autogluon-tutorial-course-v3/docs/_build/eval/tutorials/course/4b2d6613', 'lr': 0.01, 'model': 'resnet50_v1b', 'ngpus_per_trial': 1, 'nthreads_per_trial': 128, 'num_trials': 1, 'num_workers': 8, 'search_strategy': 'random', 'seed': 342, 'time_limits': 7200, 'wall_clock_tick': 1614636981.7992656}, 'total_time': 586.5652737617493, 'train_acc': 0.632425925925926, 'valid_acc': 0.9055}
print(classifier.fit_summary())
{'train_acc': 0.632425925925926, 'valid_acc': 0.9055, 'total_time': 586.5652737617493, 'best_config': {'model': 'resnet50_v1b', 'lr': 0.01, 'num_trials': 1, 'epochs': 1, 'batch_size': 16, 'nthreads_per_trial': 128, 'ngpus_per_trial': 1, 'time_limits': 7200, 'search_strategy': 'random', 'dist_ip_addrs': None, 'log_dir': '/var/lib/jenkins/workspace/workspace/autogluon-tutorial-course-v3/docs/_build/eval/tutorials/course/4b2d6613', 'custom_net': MyCifarResNet(
(features): HybridSequential(
(0): Conv2D(None -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): HybridSequential(
(0): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(1): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(2): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
)
(3): HybridSequential(
(0): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(16 -> 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
(downsample): HybridSequential(
(0): Conv2D(16 -> 32, kernel_size=(1, 1), stride=(2, 2), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(1): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(2): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(3): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(4): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(5): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(6): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(7): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
)
(4): HybridSequential(
(0): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 64, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(64 -> 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
(downsample): HybridSequential(
(0): Conv2D(32 -> 64, kernel_size=(1, 1), stride=(2, 2), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
)
(5): GlobalAvgPool2D(size=(1, 1), stride=(1, 1), padding=(0, 0), ceil_mode=True, global_pool=True, pool_type=avg, layout=NCHW)
)
(output): Dense(64 -> 10, linear)
), 'custom_optimizer': <__main__.Adam object at 0x7f20c9872050>, 'num_workers': 8, 'gpus': [0], 'seed': 342, 'final_fit': False, 'estimator': <class 'gluoncv.auto.estimators.image_classification.image_classification.ImageClassificationEstimator'>, 'wall_clock_tick': 1614636981.7992656}, 'fit_history': {'train_acc': 0.632425925925926, 'valid_acc': 0.9055, 'total_time': 586.5652737617493, 'best_config': {'model': 'resnet50_v1b', 'lr': 0.01, 'num_trials': 1, 'epochs': 1, 'batch_size': 16, 'nthreads_per_trial': 128, 'ngpus_per_trial': 1, 'time_limits': 7200, 'search_strategy': 'random', 'dist_ip_addrs': None, 'log_dir': '/var/lib/jenkins/workspace/workspace/autogluon-tutorial-course-v3/docs/_build/eval/tutorials/course/4b2d6613', 'custom_net': MyCifarResNet(
(features): HybridSequential(
(0): Conv2D(None -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): HybridSequential(
(0): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(1): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(2): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(16 -> 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
)
(3): HybridSequential(
(0): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(16 -> 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
(downsample): HybridSequential(
(0): Conv2D(16 -> 32, kernel_size=(1, 1), stride=(2, 2), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(1): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(2): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(3): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(4): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(5): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(6): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
(7): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(32 -> 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
)
(4): HybridSequential(
(0): CIFARBasicBlockV1(
(body): HybridSequential(
(0): Conv2D(32 -> 64, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
(2): Activation(relu)
(3): Conv2D(64 -> 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(4): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
(downsample): HybridSequential(
(0): Conv2D(32 -> 64, kernel_size=(1, 1), stride=(2, 2), bias=False)
(1): BatchNorm(axis=1, eps=1e-05, momentum=0.9, fix_gamma=False, use_global_stats=False, in_channels=None)
)
)
)
(5): GlobalAvgPool2D(size=(1, 1), stride=(1, 1), padding=(0, 0), ceil_mode=True, global_pool=True, pool_type=avg, layout=NCHW)
)
(output): Dense(64 -> 10, linear)
), 'custom_optimizer': <__main__.Adam object at 0x7f20c9872050>, 'num_workers': 8, 'gpus': [0], 'seed': 342, 'final_fit': False, 'estimator': <class 'gluoncv.auto.estimators.image_classification.image_classification.ImageClassificationEstimator'>, 'wall_clock_tick': 1614636981.7992656}}}