kernel_id
int64 24.2k
23.3M
| prompt
stringlengths 8
1.85M
| completetion
stringlengths 1
182k
| comp_name
stringlengths 5
57
|
|---|---|---|---|
11,896,796 | women_count = 0
women_survived_count = 0
for idx, row in train_data.iterrows() :
if row['Sex'] == 'female':
women_count += 1
if row['Survived'] == 1:
women_survived_count += 1
women_survived_count / women_count<define_variables> | GCS_DS_PATH = KaggleDatasets().get_gcs_path('tpu-getting-started')
GCS_PATH_SELECT = {
192: GCS_DS_PATH + '/tfrecords-jpeg-192x192',
224: GCS_DS_PATH + '/tfrecords-jpeg-224x224',
331: GCS_DS_PATH + '/tfrecords-jpeg-331x331',
512: GCS_DS_PATH + '/tfrecords-jpeg-512x512'
}
GCS_PATH = GCS_PATH_SELECT[IMAGE_SIZE[0]]
TRAINING_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/train/*.tfrec')+ tf.io.gfile.glob(GCS_PATH + '/val/*.tfrec')
TEST_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/test/*.tfrec' ) | Petals to the Metal - Flower Classification on TPU |
11,896,796 | predictions = []
for idx, row in test_data.iterrows() :
if row['Sex'] == 'female':
if row['Pclass'] == '1':
predictions.append(1)
elif row['Pclass'] == '2':
if row['Embarked'] == 'C':
predictions.append(1)
else:
predictions.append(0)
else:
if row['Embarked'] == 'C' or row['Fare'] > 8:
predictions.append(1)
else:
predictions.append(0)
else:
if row['Age'] < 10 and row['Pclass'] != '3':
predictions.append(1)
else:
predictions.append(0 )<feature_engineering> | CLASSES = ['pink primrose', 'hard-leaved pocket orchid', 'canterbury bells', 'sweet pea', 'wild geranium', 'tiger lily', 'moon orchid', 'bird of paradise', 'monkshood', 'globe thistle',
'snapdragon', "colt's foot", 'king protea', 'spear thistle', 'yellow iris', 'globe-flower', 'purple coneflower', 'peruvian lily', 'balloon flower', 'giant white arum lily',
'fire lily', 'pincushion flower', 'fritillary', 'red ginger', 'grape hyacinth', 'corn poppy', 'prince of wales feathers', 'stemless gentian', 'artichoke', 'sweet william',
'carnation', 'garden phlox', 'love in the mist', 'cosmos', 'alpine sea holly', 'ruby-lipped cattleya', 'cape flower', 'great masterwort', 'siam tulip', 'lenten rose',
'barberton daisy', 'daffodil', 'sword lily', 'poinsettia', 'bolero deep blue', 'wallflower', 'marigold', 'buttercup', 'daisy', 'common dandelion',
'petunia', 'wild pansy', 'primula', 'sunflower', 'lilac hibiscus', 'bishop of llandaff', 'gaura', 'geranium', 'orange dahlia', 'pink-yellow dahlia',
'cautleya spicata', 'japanese anemone', 'black-eyed susan', 'silverbush', 'californian poppy', 'osteospermum', 'spring crocus', 'iris', 'windflower', 'tree poppy',
'gazania', 'azalea', 'water lily', 'rose', 'thorn apple', 'morning glory', 'passion flower', 'lotus', 'toad lily', 'anthurium',
'frangipani', 'clematis', 'hibiscus', 'columbine', 'desert-rose', 'tree mallow', 'magnolia', 'cyclamen ', 'watercress', 'canna lily',
'hippeastrum ', 'bee balm', 'pink quill', 'foxglove', 'bougainvillea', 'camellia', 'mallow', 'mexican petunia', 'bromelia', 'blanket flower',
'trumpet creeper', 'blackberry lily', 'common tulip', 'wild rose'] | Petals to the Metal - Flower Classification on TPU |
11,896,796 | test_data['Survived'] = predictions<save_to_csv> | LR_START = 0.00001
LR_MAX = 0.00005 * strategy.num_replicas_in_sync
LR_MIN = 0.00001
LR_RAMPUP_EPOCHS = 5
LR_SUSTAIN_EPOCHS = 0
LR_EXP_DECAY =.8
def lrfn(epoch):
if epoch < LR_RAMPUP_EPOCHS:
lr =(LR_MAX - LR_START)/ LR_RAMPUP_EPOCHS * epoch + LR_START
elif epoch < LR_RAMPUP_EPOCHS + LR_SUSTAIN_EPOCHS:
lr = LR_MAX
else:
lr =(LR_MAX - LR_MIN)* LR_EXP_DECAY**(epoch - LR_RAMPUP_EPOCHS - LR_SUSTAIN_EPOCHS)+ LR_MIN
return lr
lr_callback = tf.keras.callbacks.LearningRateScheduler(lrfn, verbose = True)
rng = [i for i in range(50 if EPOCHS<50 else EPOCHS)]
y = [lrfn(x)for x in rng]
plt.plot(rng, y)
print("Learning rate schedule: {:.3g} to {:.3g} to {:.3g}".format(y[0], max(y), y[-1])) | Petals to the Metal - Flower Classification on TPU |
11,896,796 | test_data[['PassengerId', 'Survived']].to_csv('submission.csv', index=False )<set_options> | def decode_image(image_data):
image = tf.image.decode_jpeg(image_data, channels=3)
image = tf.cast(image, tf.float32)/ 255.0
image = tf.reshape(image, [*IMAGE_SIZE, 3])
return image
def read_labeled_tfrecord(example):
LABELED_TFREC_FORMAT = {
"image": tf.io.FixedLenFeature([], tf.string),
"class": tf.io.FixedLenFeature([], tf.int64),
}
example = tf.io.parse_single_example(example, LABELED_TFREC_FORMAT)
image = decode_image(example['image'])
label = tf.cast(example['class'], tf.int32)
return image, label
def read_unlabeled_tfrecord(example):
UNLABELED_TFREC_FORMAT = {
"image": tf.io.FixedLenFeature([], tf.string),
"id": tf.io.FixedLenFeature([], tf.string),
}
example = tf.io.parse_single_example(example, UNLABELED_TFREC_FORMAT)
image = decode_image(example['image'])
idnum = example['id']
return image, idnum
def load_dataset(filenames, labeled = True, ordered = False):
ignore_order = tf.data.Options()
if not ordered:
ignore_order.experimental_deterministic = False
dataset = tf.data.TFRecordDataset(filenames, num_parallel_reads = AUTO)
dataset = dataset.with_options(ignore_order)
dataset = dataset.map(read_labeled_tfrecord if labeled else read_unlabeled_tfrecord, num_parallel_calls = AUTO)
return dataset
def data_augment(image, label):
image = tf.image.random_flip_left_right(image)
return image, label
def get_training_dataset(dataset,do_aug=True):
dataset = dataset.map(data_augment, num_parallel_calls=AUTO)
if do_aug: dataset = dataset.map(transform, num_parallel_calls=AUTO)
dataset = dataset.repeat()
dataset = dataset.shuffle(2048)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.prefetch(AUTO)
return dataset
def get_validation_dataset(dataset):
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.cache()
dataset = dataset.prefetch(AUTO)
return dataset
def get_test_dataset(ordered=False):
dataset = load_dataset(TEST_FILENAMES, labeled=False, ordered=ordered)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.prefetch(AUTO)
return dataset
def count_data_items(filenames):
n = [int(re.compile(r"-([0-9]*)\." ).search(filename ).group(1)) for filename in filenames]
return np.sum(n)
NUM_TRAINING_IMAGES = int(count_data_items(TRAINING_FILENAMES)*(FOLDS-1.) /FOLDS)
NUM_VALIDATION_IMAGES = int(count_data_items(TRAINING_FILENAMES)*(1./FOLDS))
NUM_TEST_IMAGES = count_data_items(TEST_FILENAMES)
STEPS_PER_EPOCH = NUM_TRAINING_IMAGES // BATCH_SIZE
print('Dataset: {} training images, {} validation images, {} unlabeled test images'.format(NUM_TRAINING_IMAGES, NUM_VALIDATION_IMAGES, NUM_TEST_IMAGES)) | Petals to the Metal - Flower Classification on TPU |
11,896,796 | %matplotlib inline
<load_from_csv> | def get_mat(rotation, shear, height_zoom, width_zoom, height_shift, width_shift):
rotation = math.pi * rotation / 180.
shear = math.pi * shear / 180.
c1 = tf.math.cos(rotation)
s1 = tf.math.sin(rotation)
one = tf.constant([1],dtype='float32')
zero = tf.constant([0],dtype='float32')
rotation_matrix = tf.reshape(tf.concat([c1,s1,zero, -s1,c1,zero, zero,zero,one],axis=0),[3,3])
c2 = tf.math.cos(shear)
s2 = tf.math.sin(shear)
shear_matrix = tf.reshape(tf.concat([one,s2,zero, zero,c2,zero, zero,zero,one],axis=0),[3,3])
zoom_matrix = tf.reshape(tf.concat([one/height_zoom,zero,zero, zero,one/width_zoom,zero, zero,zero,one],axis=0),[3,3])
shift_matrix = tf.reshape(tf.concat([one,zero,height_shift, zero,one,width_shift, zero,zero,one],axis=0),[3,3])
return K.dot(K.dot(rotation_matrix, shear_matrix), K.dot(zoom_matrix, shift_matrix)) | Petals to the Metal - Flower Classification on TPU |
11,896,796 | test_df = pd.read_csv("/kaggle/input/titanic/test.csv")
train_df = pd.read_csv("/kaggle/input/titanic/train.csv" )<sort_values> | def transform(image,label):
DIM = IMAGE_SIZE[0]
XDIM = DIM%2
rot = 15.* tf.random.normal([1],dtype='float32')
shr = 5.* tf.random.normal([1],dtype='float32')
h_zoom = 1.0 + tf.random.normal([1],dtype='float32')/10.
w_zoom = 1.0 + tf.random.normal([1],dtype='float32')/10.
h_shift = 16.* tf.random.normal([1],dtype='float32')
w_shift = 16.* tf.random.normal([1],dtype='float32')
m = get_mat(rot,shr,h_zoom,w_zoom,h_shift,w_shift)
x = tf.repeat(tf.range(DIM//2,-DIM//2,-1), DIM)
y = tf.tile(tf.range(-DIM//2,DIM//2),[DIM])
z = tf.ones([DIM*DIM],dtype='int32')
idx = tf.stack([x,y,z])
idx2 = K.dot(m,tf.cast(idx,dtype='float32'))
idx2 = K.cast(idx2,dtype='int32')
idx2 = K.clip(idx2,-DIM//2+XDIM+1,DIM//2)
idx3 = tf.stack([DIM//2-idx2[0,], DIM//2-1+idx2[1,]])
d = tf.gather_nd(image,tf.transpose(idx3))
return tf.reshape(d,[DIM,DIM,3]),label
| Petals to the Metal - Flower Classification on TPU |
11,896,796 | total = train_df.isnull().sum().sort_values(ascending=False)
percent_1 = train_df.isnull().sum() /train_df.isnull().count() *100
percent_2 =(round(percent_1, 1)).sort_values(ascending=False)
missing_data = pd.concat([total, percent_2], axis=1, keys=['Total', '%'])
missing_data.head(5 )<data_type_conversions> | def get_model() :
with strategy.scope() :
rnet = InceptionResNetV2(
input_shape=(IMAGE_SIZE[0], IMAGE_SIZE[1], 3),
weights='imagenet',
include_top=False
)
rnet.trainable = True
model = tf.keras.Sequential([
rnet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation='softmax',dtype='float32')
])
model.compile(
optimizer='adam',
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
return model
def train_cross_validate(folds = 5):
histories = []
models = []
early_stopping = tf.keras.callbacks.EarlyStopping(monitor = 'val_loss', patience = 3)
kfold = KFold(folds, shuffle = True, random_state = SEED)
for f,(trn_ind, val_ind)in enumerate(kfold.split(TRAINING_FILENAMES)) :
print() ; print('
print('
print('
train_dataset = load_dataset(list(pd.DataFrame({'TRAINING_FILENAMES': TRAINING_FILENAMES} ).loc[trn_ind]['TRAINING_FILENAMES']), labeled = True)
val_dataset = load_dataset(list(pd.DataFrame({'TRAINING_FILENAMES': TRAINING_FILENAMES} ).loc[val_ind]['TRAINING_FILENAMES']), labeled = True, ordered = True)
model = get_model()
history = model.fit(
get_training_dataset(train_dataset),
steps_per_epoch = STEPS_PER_EPOCH,
epochs = EPOCHS,
callbacks = [lr_callback],
validation_data = get_validation_dataset(val_dataset),
verbose=2
)
models.append(model)
histories.append(history)
return histories, models
def train_and_predict(folds = 5):
test_ds = get_test_dataset(ordered=True)
test_images_ds = test_ds.map(lambda image, idnum: image)
print('Start training %i folds'%folds)
histories, models = train_cross_validate(folds = folds)
print('Computing predictions...')
probabilities = np.average([models[i].predict(test_images_ds)for i in range(folds)], axis = 0)
predictions = np.argmax(probabilities, axis=-1)
print('Generating submission.csv file...')
test_ids_ds = test_ds.map(lambda image, idnum: idnum ).unbatch()
test_ids = next(iter(test_ids_ds.batch(NUM_TEST_IMAGES)) ).numpy().astype('U')
np.savetxt('submission.csv', np.rec.fromarrays([test_ids, predictions]), fmt=['%s', '%d'], delimiter=',', header='id,label', comments='')
return histories, models
histories_inception, inception_model = train_and_predict(folds = FOLDS ) | Petals to the Metal - Flower Classification on TPU |
11,896,796 | data = [train_df, test_df]
for dataset in data:
dataset['relatives'] = dataset['SibSp'] + dataset['Parch']
dataset.loc[dataset['relatives'] > 0, 'not_alone'] = 0
dataset.loc[dataset['relatives'] == 0, 'not_alone'] = 1
dataset['not_alone'] = dataset['not_alone'].astype(int)
train_df['not_alone'].value_counts()<drop_column> | def get_model() :
with strategy.scope() :
rnet = NASNetLarge(
input_shape=(IMAGE_SIZE[0], IMAGE_SIZE[1], 3),
weights='imagenet',
include_top=False
)
rnet.trainable = True
model = tf.keras.Sequential([
rnet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation='softmax',dtype='float32')
])
model.compile(
optimizer='adam',
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
return model
def train_cross_validate(folds = 5):
histories = []
models = []
early_stopping = tf.keras.callbacks.EarlyStopping(monitor = 'val_loss', patience = 3)
kfold = KFold(folds, shuffle = True, random_state = SEED)
for f,(trn_ind, val_ind)in enumerate(kfold.split(TRAINING_FILENAMES)) :
print() ; print('
print('
print('
train_dataset = load_dataset(list(pd.DataFrame({'TRAINING_FILENAMES': TRAINING_FILENAMES} ).loc[trn_ind]['TRAINING_FILENAMES']), labeled = True)
val_dataset = load_dataset(list(pd.DataFrame({'TRAINING_FILENAMES': TRAINING_FILENAMES} ).loc[val_ind]['TRAINING_FILENAMES']), labeled = True, ordered = True)
model = get_model()
history = model.fit(
get_training_dataset(train_dataset),
steps_per_epoch = STEPS_PER_EPOCH,
epochs = EPOCHS,
callbacks = [lr_callback],
validation_data = get_validation_dataset(val_dataset),
verbose=2
)
models.append(model)
histories.append(history)
return histories, models
def train_and_predict(folds = 5):
test_ds = get_test_dataset(ordered=True)
test_images_ds = test_ds.map(lambda image, idnum: image)
print('Start training %i folds'%folds)
histories, models = train_cross_validate(folds = folds)
print('Computing predictions...')
probabilities = np.average([models[i].predict(test_images_ds)for i in range(folds)], axis = 0)
predictions = np.argmax(probabilities, axis=-1)
print('Generating submission.csv file...')
test_ids_ds = test_ds.map(lambda image, idnum: idnum ).unbatch()
test_ids = next(iter(test_ids_ds.batch(NUM_TEST_IMAGES)) ).numpy().astype('U')
np.savetxt('submission.csv', np.rec.fromarrays([test_ids, predictions]), fmt=['%s', '%d'], delimiter=',', header='id,label', comments='')
return histories, models
| Petals to the Metal - Flower Classification on TPU |
11,896,796 | train_df = train_df.drop(['PassengerId'], axis=1 )<categorify> | %%time
all_labels = []; all_prob = []; all_pred = []
kfold = KFold(FOLDS, shuffle = True, random_state = SEED)
for j,(trn_ind, val_ind)in enumerate(kfold.split(TRAINING_FILENAMES)) :
print('Inferring fold',j+1,'validation images...')
VAL_FILES = list(pd.DataFrame({'TRAINING_FILENAMES': TRAINING_FILENAMES} ).loc[val_ind]['TRAINING_FILENAMES'])
NUM_VALIDATION_IMAGES = count_data_items(VAL_FILES)
cmdataset = get_validation_dataset(load_dataset(VAL_FILES, labeled = True, ordered = True))
images_ds = cmdataset.map(lambda image, label: image)
labels_ds = cmdataset.map(lambda image, label: label ).unbatch()
all_labels.append(next(iter(labels_ds.batch(NUM_VALIDATION_IMAGES)) ).numpy())
prob = inception_model[j].predict(images_ds)
all_prob.append(prob)
all_pred.append(np.argmax(prob, axis=-1))
cm_correct_labels = np.concatenate(all_labels)
cm_probabilities = np.concatenate(all_prob)
cm_predictions = np.concatenate(all_pred ) | Petals to the Metal - Flower Classification on TPU |
11,896,796 | deck = {"A": 1, "B": 2, "C": 3, "D": 4, "E": 5, "F": 6, "G": 7, "U": 8}
data = [train_df, test_df]
for dataset in data:
dataset['Cabin'] = dataset['Cabin'].fillna("U0")
dataset['Deck'] = dataset['Cabin'].map(lambda x: re.compile("([a-zA-Z]+)" ).search(x ).group())
dataset['Deck'] = dataset['Deck'].map(deck)
dataset['Deck'] = dataset['Deck'].fillna(0)
dataset['Deck'] = dataset['Deck'].astype(int)
train_df = train_df.drop(['Cabin'], axis=1)
test_df = test_df.drop(['Cabin'], axis=1 )<data_type_conversions> | cmat = confusion_matrix(cm_correct_labels, cm_predictions, labels=range(len(CLASSES)))
score = f1_score(cm_correct_labels, cm_predictions, labels=range(len(CLASSES)) , average='macro')
precision = precision_score(cm_correct_labels, cm_predictions, labels=range(len(CLASSES)) , average='macro')
recall = recall_score(cm_correct_labels, cm_predictions, labels=range(len(CLASSES)) , average='macro')
display_confusion_matrix(cmat, score, precision, recall)
print('f1 score: {:.3f}, precision: {:.3f}, recall: {:.3f}'.format(score, precision, recall)) ; print() | Petals to the Metal - Flower Classification on TPU |
11,896,796 | data = [train_df, test_df]
for dataset in data:
mean = train_df["Age"].mean()
std = test_df["Age"].std()
is_null = dataset["Age"].isnull().sum()
rand_age = np.random.randint(mean - std, mean + std, size = is_null)
age_slice = dataset["Age"].copy()
age_slice[np.isnan(age_slice)] = rand_age
dataset["Age"] = age_slice
dataset["Age"] = train_df["Age"].astype(int)
train_df["Age"].isnull().sum()<categorify> | Petals to the Metal - Flower Classification on TPU |
|
11,756,240 | common_value = 'S'
data = [train_df, test_df]
for dataset in data:
dataset['Embarked'] = dataset['Embarked'].fillna(common_value )<data_type_conversions> | print("TF version " + tf.__version__)
AUTO = tf.data.experimental.AUTOTUNE | Petals to the Metal - Flower Classification on TPU |
11,756,240 | data = [train_df, test_df]
for dataset in data:
dataset['Fare'] = dataset['Fare'].fillna(0)
dataset['Fare'] = dataset['Fare'].astype(int)
data = [train_df, test_df]
titles = {"Mr": 1, "Miss": 2, "Mrs": 3, "Master": 4, "Rare": 5}
for dataset in data:
dataset['Title'] = dataset.Name.str.extract('([A-Za-z]+)\.', expand=False)
dataset['Title'] = dataset['Title'].replace(['Lady', 'Countess','Capt', 'Col','Don', 'Dr',\
'Major', 'Rev', 'Sir', 'Jonkheer', 'Dona'], 'Rare')
dataset['Title'] = dataset['Title'].replace('Mlle', 'Miss')
dataset['Title'] = dataset['Title'].replace('Ms', 'Miss')
dataset['Title'] = dataset['Title'].replace('Mme', 'Mrs')
dataset['Title'] = dataset['Title'].map(titles)
dataset['Title'] = dataset['Title'].fillna(0)
train_df = train_df.drop(['Name'], axis=1)
test_df = test_df.drop(['Name'], axis=1)
genders = {"male": 0, "female": 1}
data = [train_df, test_df]
for dataset in data:
dataset['Sex'] = dataset['Sex'].map(genders)
train_df['Ticket'].describe()
train_df = train_df.drop(['Ticket'], axis=1)
test_df = test_df.drop(['Ticket'], axis=1 )<categorify> | try:
tpu = tf.distribute.cluster_resolver.TPUClusterResolver()
print('Running on TPU ', tpu.master())
except ValueError:
tpu = None
if tpu:
tf.config.experimental_connect_to_cluster(tpu)
tf.tpu.experimental.initialize_tpu_system(tpu)
strategy = tf.distribute.experimental.TPUStrategy(tpu)
else:
strategy = tf.distribute.get_strategy()
print("REPLICAS: ", strategy.num_replicas_in_sync ) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | ports = {"S": 0, "C": 1, "Q": 2}
data = [train_df, test_df]
for dataset in data:
dataset['Embarked'] = dataset['Embarked'].map(ports )<feature_engineering> | IMAGE_SIZE = [512, 512]
EPOCHS = 30
BATCH_SIZE = 16 * strategy.num_replicas_in_sync
SKIP_VALIDATION = False
TTA_NUM = 5
SEED = 42
random.seed(SEED)
np.random.seed(SEED)
tf.random.set_seed(SEED ) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | data = [train_df, test_df]
for dataset in data:
dataset['Age'] = dataset['Age'].astype(int)
dataset.loc[ dataset['Age'] <= 11, 'Age'] = 0
dataset.loc[(dataset['Age'] > 11)&(dataset['Age'] <= 18), 'Age'] = 1
dataset.loc[(dataset['Age'] > 18)&(dataset['Age'] <= 22), 'Age'] = 2
dataset.loc[(dataset['Age'] > 22)&(dataset['Age'] <= 27), 'Age'] = 3
dataset.loc[(dataset['Age'] > 27)&(dataset['Age'] <= 33), 'Age'] = 4
dataset.loc[(dataset['Age'] > 33)&(dataset['Age'] <= 40), 'Age'] = 5
dataset.loc[(dataset['Age'] > 40)&(dataset['Age'] <= 66), 'Age'] = 6
dataset.loc[ dataset['Age'] > 66, 'Age'] = 6
<feature_engineering> | GCS_DS_PATH = KaggleDatasets().get_gcs_path('tpu-getting-started')
GCS_PATH_SELECT = {
192: GCS_DS_PATH + '/tfrecords-jpeg-192x192',
224: GCS_DS_PATH + '/tfrecords-jpeg-224x224',
331: GCS_DS_PATH + '/tfrecords-jpeg-331x331',
512: GCS_DS_PATH + '/tfrecords-jpeg-512x512'
}
GCS_PATH = GCS_PATH_SELECT[IMAGE_SIZE[0]]
| Petals to the Metal - Flower Classification on TPU |
11,756,240 | data = [train_df, test_df]
for dataset in data:
dataset.loc[ dataset['Fare'] <= 7.91, 'Fare'] = 0
dataset.loc[(dataset['Fare'] > 7.91)&(dataset['Fare'] <= 14.454), 'Fare'] = 1
dataset.loc[(dataset['Fare'] > 14.454)&(dataset['Fare'] <= 31), 'Fare'] = 2
dataset.loc[(dataset['Fare'] > 31)&(dataset['Fare'] <= 99), 'Fare'] = 3
dataset.loc[(dataset['Fare'] > 99)&(dataset['Fare'] <= 250), 'Fare'] = 4
dataset.loc[ dataset['Fare'] > 250, 'Fare'] = 5
dataset['Fare'] = dataset['Fare'].astype(int )<feature_engineering> | TRAINING_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/train/*.tfrec')
VALIDATION_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/val/*.tfrec')
TEST_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/test/*.tfrec' ) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | data = [train_df, test_df]
for dataset in data:
dataset['Age_Class']= dataset['Age']* dataset['Pclass']<data_type_conversions> | np.set_printoptions(threshold=15, linewidth=80)
def batch_to_numpy_images_and_labels(data):
images, labels = data
numpy_images = images.numpy()
numpy_labels = labels.numpy()
if numpy_labels.dtype == object:
numpy_labels = [None for _ in enumerate(numpy_images)]
return numpy_images, numpy_labels
def title_from_label_and_target(label, correct_label):
if correct_label is None:
return CLASSES[label], True
correct =(label == correct_label)
return "{} [{}{}{}]".format(CLASSES[label], 'OK' if correct else 'NO', u"\u2192" if not correct else '',
CLASSES[correct_label] if not correct else ''), correct
| Petals to the Metal - Flower Classification on TPU |
11,756,240 | for dataset in data:
dataset['Fare_Per_Person'] = dataset['Fare']/(dataset['relatives']+1)
dataset['Fare_Per_Person'] = dataset['Fare_Per_Person'].astype(int)
train_df.head(10 )<save_to_csv> | def display_batch_of_images(databatch, predictions=None):
images, labels = batch_to_numpy_images_and_labels(databatch)
if labels is None:
labels = [None for _ in enumerate(images)]
rows = int(math.sqrt(len(images)))
cols = len(images)//rows
FIGSIZE = 13.0
SPACING = 0.1
subplot=(rows,cols,1)
if rows < cols:
plt.figure(figsize=(FIGSIZE,FIGSIZE/cols*rows))
else:
plt.figure(figsize=(FIGSIZE/rows*cols,FIGSIZE))
for i,(image, label)in enumerate(zip(images[:rows*cols], labels[:rows*cols])) :
title = '' if label is None else CLASSES[label]
correct = True
if predictions is not None:
title, correct = title_from_label_and_target(predictions[i], label)
dynamic_titlesize = FIGSIZE*SPACING/max(rows,cols)*40+3
subplot = display_one_flower(image, title, subplot, not correct, titlesize=dynamic_titlesize)
plt.tight_layout()
if label is None and predictions is None:
plt.subplots_adjust(wspace=0, hspace=0)
else:
plt.subplots_adjust(wspace=SPACING, hspace=SPACING)
plt.show() | Petals to the Metal - Flower Classification on TPU |
11,756,240 | X_train = train_df.drop("Survived", axis=1)
Y_train = train_df["Survived"]
X_test = test_df.drop("PassengerId", axis=1 ).copy()
decision_tree = DecisionTreeClassifier()
decision_tree.fit(X_train, Y_train)
predictions = decision_tree.predict(X_test)
output = pd.DataFrame({'PassengerId': test_df.PassengerId, 'Survived': predictions})
output.to_csv('my_submission.csv', index=False)
print("Your submission was successfully saved!" )<set_options> | def random_blockout(img, sl=0.1, sh=0.2, rl=0.4):
p=random.random()
if p>=0.25:
w, h, c = IMAGE_SIZE[0], IMAGE_SIZE[1], 3
origin_area = tf.cast(h*w, tf.float32)
e_size_l = tf.cast(tf.round(tf.sqrt(origin_area * sl * rl)) , tf.int32)
e_size_h = tf.cast(tf.round(tf.sqrt(origin_area * sh / rl)) , tf.int32)
e_height_h = tf.minimum(e_size_h, h)
e_width_h = tf.minimum(e_size_h, w)
erase_height = tf.random.uniform(shape=[], minval=e_size_l, maxval=e_height_h, dtype=tf.int32)
erase_width = tf.random.uniform(shape=[], minval=e_size_l, maxval=e_width_h, dtype=tf.int32)
erase_area = tf.zeros(shape=[erase_height, erase_width, c])
erase_area = tf.cast(erase_area, tf.uint8)
pad_h = h - erase_height
pad_top = tf.random.uniform(shape=[], minval=0, maxval=pad_h, dtype=tf.int32)
pad_bottom = pad_h - pad_top
pad_w = w - erase_width
pad_left = tf.random.uniform(shape=[], minval=0, maxval=pad_w, dtype=tf.int32)
pad_right = pad_w - pad_left
erase_mask = tf.pad([erase_area], [[0,0],[pad_top, pad_bottom], [pad_left, pad_right], [0,0]], constant_values=1)
erase_mask = tf.squeeze(erase_mask, axis=0)
erased_img = tf.multiply(tf.cast(img,tf.float32), tf.cast(erase_mask, tf.float32))
return tf.cast(erased_img, img.dtype)
else:
return tf.cast(img, img.dtype ) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | %matplotlib inline
np.random.seed(2)
<load_from_csv> | def decode_image(image_data):
image = tf.image.decode_jpeg(image_data, channels=3)
image = tf.cast(image, tf.float32)/ 255.0
image = tf.reshape(image, [*IMAGE_SIZE, 3])
return image
def read_labeled_tfrecord(example):
LABELED_TFREC_FORMAT = {
"image": tf.io.FixedLenFeature([], tf.string),
"class": tf.io.FixedLenFeature([], tf.int64),
}
example = tf.io.parse_single_example(example, LABELED_TFREC_FORMAT)
image = decode_image(example['image'])
label = tf.cast(example['class'], tf.int32)
return image, label
def read_unlabeled_tfrecord(example):
UNLABELED_TFREC_FORMAT = {
"image": tf.io.FixedLenFeature([], tf.string),
"id": tf.io.FixedLenFeature([], tf.string),
}
example = tf.io.parse_single_example(example, UNLABELED_TFREC_FORMAT)
image = decode_image(example['image'])
idnum = example['id']
return image, idnum | Petals to the Metal - Flower Classification on TPU |
11,756,240 | train = pd.read_csv('/kaggle/input/digit-recognizer/train.csv')
test = pd.read_csv('/kaggle/input/digit-recognizer/test.csv' )<prepare_x_and_y> | def load_dataset(filenames, labeled=True, ordered=False):
ignore_order = tf.data.Options()
if not ordered:
ignore_order.experimental_deterministic = False
dataset = tf.data.TFRecordDataset(filenames, num_parallel_reads=AUTO)
dataset = dataset.with_options(ignore_order)
dataset = dataset.map(read_labeled_tfrecord if labeled else read_unlabeled_tfrecord, num_parallel_calls=AUTO)
return dataset
| Petals to the Metal - Flower Classification on TPU |
11,756,240 | Y_train = train['label']
X_train = train.drop(labels=['label'], axis=1)
del train<count_values> | def get_training_dataset() :
dataset = load_dataset(TRAINING_FILENAMES, labeled=True)
dataset = dataset.map(data_augment, num_parallel_calls=AUTO)
dataset = dataset.repeat()
dataset = dataset.shuffle(2048)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.prefetch(AUTO)
return dataset | Petals to the Metal - Flower Classification on TPU |
11,756,240 | Y_train.value_counts()<feature_engineering> | def get_validation_dataset(ordered=False):
dataset = load_dataset(VALIDATION_FILENAMES, labeled=True, ordered=ordered)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.cache()
dataset = dataset.prefetch(AUTO)
return dataset | Petals to the Metal - Flower Classification on TPU |
11,756,240 | X_train = X_train / 255.0
test = test / 255.0<categorify> | def get_test_dataset(ordered=False):
dataset = load_dataset(TEST_FILENAMES, labeled=False, ordered=ordered)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.prefetch(AUTO)
return dataset
| Petals to the Metal - Flower Classification on TPU |
11,756,240 | Y_train = to_categorical(Y_train, num_classes=10)
print(Y_train.shape )<define_variables> | def count_data_items(filenames):
n = [int(re.compile(r"-([0-9]*)\." ).search(filename ).group(1)) for filename in filenames]
return np.sum(n ) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | random_seed = 2<split> | NUM_TRAINING_IMAGES = count_data_items(TRAINING_FILENAMES)
NUM_VALIDATION_IMAGES = count_data_items(VALIDATION_FILENAMES)
NUM_TEST_IMAGES = count_data_items(TEST_FILENAMES)
STEPS_PER_EPOCH = NUM_TRAINING_IMAGES // BATCH_SIZE
print('Dataset: {} training images, {} validation images, {} unlabeled test images'.format(NUM_TRAINING_IMAGES, NUM_VALIDATION_IMAGES, NUM_TEST_IMAGES)) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | X_train, X_val, Y_train, Y_val = train_test_split(X_train, Y_train, test_size=0.1, random_state=random_seed )<import_modules> | if SKIP_VALIDATION:
TRAINING_FILENAMES = TRAINING_FILENAMES + VALIDATION_FILENAMES | Petals to the Metal - Flower Classification on TPU |
11,756,240 | from keras.models import Sequential
from keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPool2D
from keras.optimizers import RMSprop
from keras.preprocessing.image import ImageDataGenerator
from keras.callbacks import ReduceLROnPlateau<choose_model_class> | print("Training data shapes:")
for image, label in get_training_dataset().take(3):
print(image.numpy().shape, label.numpy().shape)
print("Training data label examples:", label.numpy())
print("Validation data shapes:")
for image, label in get_validation_dataset().take(3):
print(image.numpy().shape, label.numpy().shape)
print("Validation data label examples:", label.numpy())
print("Test data shapes:")
for image, idnum in get_test_dataset().take(3):
print(image.numpy().shape, idnum.numpy().shape)
print("Test data IDs:", idnum.numpy().astype('U')) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | model = Sequential()
model.add(Conv2D(filters=32, kernel_size=(5, 5), padding='same', activation='relu', input_shape=(28, 28, 1)))
model.add(Conv2D(filters=32, kernel_size=(5, 5), padding='same', activation='relu'))
model.add(MaxPool2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Conv2D(filters=64, kernel_size=(3, 3), padding='same', activation='relu'))
model.add(Conv2D(filters=64, kernel_size=(3, 3), padding='same', activation='relu'))
model.add(MaxPool2D(pool_size=(2, 2), strides=(2, 2)))
model.add(Dropout(0.25))
model.add(Flatten())
model.add(Dense(256, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(10, activation='softmax'))<choose_model_class> | training_dataset = get_training_dataset()
training_dataset = training_dataset.unbatch().batch(20)
train_batch = iter(training_dataset ) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | optimizer = RMSprop(lr=0.001, rho=0.9, epsilon=1e-08, decay=0.0 )<choose_model_class> | CLASSES = ['pink primrose', 'hard-leaved pocket orchid', 'canterbury bells', 'sweet pea', 'wild geranium', 'tiger lily', 'moon orchid', 'bird of paradise', 'monkshood', 'globe thistle',
'snapdragon', "colt's foot", 'king protea', 'spear thistle', 'yellow iris', 'globe-flower', 'purple coneflower', 'peruvian lily', 'balloon flower', 'giant white arum lily',
'fire lily', 'pincushion flower', 'fritillary', 'red ginger', 'grape hyacinth', 'corn poppy', 'prince of wales feathers', 'stemless gentian', 'artichoke', 'sweet william',
'carnation', 'garden phlox', 'love in the mist', 'cosmos', 'alpine sea holly', 'ruby-lipped cattleya', 'cape flower', 'great masterwort', 'siam tulip', 'lenten rose',
'barberton daisy', 'daffodil', 'sword lily', 'poinsettia', 'bolero deep blue', 'wallflower', 'marigold', 'buttercup', 'daisy', 'common dandelion',
'petunia', 'wild pansy', 'primula', 'sunflower', 'lilac hibiscus', 'bishop of llandaff', 'gaura', 'geranium', 'orange dahlia', 'pink-yellow dahlia',
'cautleya spicata', 'japanese anemone', 'black-eyed susan', 'silverbush', 'californian poppy', 'osteospermum', 'spring crocus', 'iris', 'windflower', 'tree poppy',
'gazania', 'azalea', 'water lily', 'rose', 'thorn apple', 'morning glory', 'passion flower', 'lotus', 'toad lily', 'anthurium',
'frangipani', 'clematis', 'hibiscus', 'columbine', 'desert-rose', 'tree mallow', 'magnolia', 'cyclamen ', 'watercress', 'canna lily',
'hippeastrum ', 'bee balm', 'pink quill', 'foxglove', 'bougainvillea', 'camellia', 'mallow', 'mexican petunia', 'bromelia', 'blanket flower',
'trumpet creeper', 'blackberry lily', 'common tulip', 'wild rose']
| Petals to the Metal - Flower Classification on TPU |
11,756,240 | model.compile(optimizer, loss='categorical_crossentropy', metrics=['accuracy'] )<choose_model_class> | display_batch_of_images(next(train_batch)) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | learning_rate_reduction = ReduceLROnPlateau(monitor='val_accuracy',
patience=3,
verbose=1,
factor=0.5,
min_lr=0.00001 )<define_variables> | test_dataset = get_test_dataset()
test_dataset = test_dataset.unbatch().batch(20)
test_batch = iter(test_dataset ) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | epochs = 30
batch_size = 86<train_model> | display_batch_of_images(next(test_batch)) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | datagen = ImageDataGenerator(
featurewise_center=False,
samplewise_center=False,
featurewise_std_normalization=False,
samplewise_std_normalization=False,
zca_whitening=False,
rotation_range=10,
zoom_range=0.1,
height_shift_range=0.1,
horizontal_flip=False,
vertical_flip=False
)
datagen.fit(X_train )<train_model> | LR_START = 0.00001
LR_MAX = 0.00005 * strategy.num_replicas_in_sync
LR_MIN = 0.00001
LR_RAMPUP_EPOCHS = 5
LR_SUSTAIN_EPOCHS = 0
LR_EXP_DECAY =.8
def lrfn(epoch):
if epoch < LR_RAMPUP_EPOCHS:
lr =(LR_MAX - LR_START)/ LR_RAMPUP_EPOCHS * epoch + LR_START
elif epoch < LR_RAMPUP_EPOCHS + LR_SUSTAIN_EPOCHS:
lr = LR_MAX
else:
lr =(LR_MAX - LR_MIN)* LR_EXP_DECAY**(epoch - LR_RAMPUP_EPOCHS - LR_SUSTAIN_EPOCHS)+ LR_MIN
return lr
lr_callback = tf.keras.callbacks.LearningRateScheduler(lrfn, verbose = True)
rng = [i for i in range(25 if EPOCHS<25 else EPOCHS)]
y = [lrfn(x)for x in rng]
plt.plot(rng, y)
print("Learning rate schedule: {:.3g} to {:.3g} to {:.3g}".format(y[0], max(y), y[-1])) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | history = model.fit_generator(datagen.flow(X_train, Y_train, batch_size=batch_size),
epochs=epochs,
validation_data=(X_val, Y_val),
verbose=2,
steps_per_epoch=X_train.shape[0] // batch_size,
callbacks=[learning_rate_reduction] )<import_modules> | !pip install -q efficientnet
| Petals to the Metal - Flower Classification on TPU |
11,756,240 | import itertools
from sklearn.metrics import confusion_matrix<compute_test_metric> | with strategy.scope() :
enet = efn.EfficientNetB7(input_shape=[*IMAGE_SIZE, 3], weights='noisy-student', include_top=False)
enet.trainable = True
model1 = tf.keras.Sequential([
enet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation='softmax')
] ) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | errors =(Y_pred_classes - Y_true)!= 0
print(errors)
print(np.sum(errors))<define_variables> | model1.compile(
optimizer=tf.keras.optimizers.Adam() ,
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
| Petals to the Metal - Flower Classification on TPU |
11,756,240 | Y_pred_classes_errors = Y_pred_classes[errors]
Y_pred_errors = Y_pred[errors]
Y_true_errors = Y_true[errors]
X_val_errors = X_val[errors]<predict_on_test> | if not SKIP_VALIDATION:
history1 = model1.fit(get_training_dataset() ,
steps_per_epoch=STEPS_PER_EPOCH,
epochs=EPOCHS,
validation_data=get_validation_dataset() ,
callbacks = [lr_callback])
else:
history1 = model1.fit(get_training_dataset() ,
steps_per_epoch=STEPS_PER_EPOCH,
epochs=EPOCHS,
callbacks = [lr_callback] ) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | results = model.predict(test)
results.shape<concatenate> | with strategy.scope() :
densenet = tf.keras.applications.DenseNet201(input_shape=[*IMAGE_SIZE, 3], weights='imagenet', include_top=False)
densenet.trainable = True
model2 = tf.keras.Sequential([
densenet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation='softmax')
])
model2.compile(
optimizer=tf.keras.optimizers.Adam() ,
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
model2.summary() | Petals to the Metal - Flower Classification on TPU |
11,756,240 | submission = pd.concat([pd.Series(range(1, 28001), name='ImageId'), results], axis=1)
submission.head()<save_to_csv> | if not SKIP_VALIDATION:
history2 = model2.fit(get_training_dataset() ,
steps_per_epoch=STEPS_PER_EPOCH,
epochs=EPOCHS,
validation_data=get_validation_dataset() ,
callbacks = [lr_callback])
else:
history2 = model2.fit(get_training_dataset() ,
steps_per_epoch=STEPS_PER_EPOCH,
epochs=EPOCHS,
callbacks = [lr_callback] ) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | submission.to_csv('cnn_mnist_augmentation.csv', index=False )<set_options> | if not SKIP_VALIDATION:
cmdataset = get_validation_dataset(ordered=True)
images_ds = cmdataset.map(lambda image, label: image)
labels_ds = cmdataset.map(lambda image, label: label ).unbatch()
cm_correct_labels = next(iter(labels_ds.batch(NUM_VALIDATION_IMAGES)) ).numpy()
m1 = model1.predict(images_ds)
m2 = model2.predict(images_ds)
scores = []
for alpha in np.linspace(0,1,100):
cm_probabilities = alpha*m1+(1-alpha)*m2
cm_predictions = np.argmax(cm_probabilities, axis=-1)
scores.append(f1_score(cm_correct_labels, cm_predictions, labels=range(len(CLASSES)) , average='macro'))
best_alpha = np.argmax(scores)/100
else:
best_alpha = 0.51
print('Best alpha: ' + str(best_alpha)) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | %matplotlib inline
PUNCT_TO_REMOVE = string.punctuation
STOPWORDS = set(stopwords.words('english'))
stemmer = PorterStemmer()<load_from_csv> | if not SKIP_VALIDATION:
cm_probabilities = best_alpha*m1 +(1-best_alpha)*m2
cm_predictions = np.argmax(cm_probabilities, axis=-1)
cmat = confusion_matrix(cm_correct_labels, cm_predictions, labels=range(len(CLASSES)))
score = f1_score(cm_correct_labels, cm_predictions, labels=range(len(CLASSES)) , average='macro')
precision = precision_score(cm_correct_labels, cm_predictions, labels=range(len(CLASSES)) , average='macro')
recall = recall_score(cm_correct_labels, cm_predictions, labels=range(len(CLASSES)) , average='macro')
display_confusion_matrix(cmat, score, precision, recall)
print('f1 score: {:.3f}, precision: {:.3f}, recall: {:.3f}'.format(score, precision, recall)) | Petals to the Metal - Flower Classification on TPU |
11,756,240 | df_train = pd.read_csv('.. /input/nlp-getting-started/train.csv')
df_test = pd.read_csv('.. /input/nlp-getting-started/test.csv' )<string_transform> | def predict_tta(model, n_iter):
probs = []
for i in range(n_iter):
test_ds = get_test_dataset(ordered=True)
test_images_ds = test_ds.map(lambda image, idnum: image)
probs.append(model.predict(test_images_ds,verbose=0))
return probs | Petals to the Metal - Flower Classification on TPU |
11,756,240 | <feature_engineering><EOS> | test_ds = get_test_dataset(ordered=True)
print('Calculating predictions...')
test_images_ds = test_ds.map(lambda image, idnum: image)
probs1 = np.mean(predict_tta(model1, TTA_NUM), axis=0)
probs2 = np.mean(predict_tta(model2, TTA_NUM), axis=0)
probabilities = best_alpha*probs1 +(1-best_alpha)*probs2
predictions = np.argmax(probabilities, axis=-1)
print('Generating submission file...')
test_ids_ds = test_ds.map(lambda image, idnum: idnum ).unbatch()
test_ids = next(iter(test_ids_ds.batch(NUM_TEST_IMAGES)) ).numpy().astype('U')
np.savetxt('submission.csv', np.rec.fromarrays([test_ids, predictions]), fmt=['%s', '%d'], delimiter=',', header='id,label', comments='' ) | Petals to the Metal - Flower Classification on TPU |
11,521,017 | <SOS> metric: macrofscore Kaggle data source: petals-to-the-metal-flower-classification-on-tpu<prepare_x_and_y> | print('Tensorflow version ' + tf.__version__)
!pip install -q efficientnet
| Petals to the Metal - Flower Classification on TPU |
11,521,017 | X = df_train['text_processed']
y = df_train['target']<split> | try:
tpu = tf.distribute.cluster_resolver.TPUClusterResolver()
print('Running on TPU ', tpu.master())
except ValueError:
tpu = None
if tpu:
tf.config.experimental_connect_to_cluster(tpu)
tf.tpu.experimental.initialize_tpu_system(tpu)
strategy = tf.distribute.experimental.TPUStrategy(tpu)
else:
strategy = tf.distribute.get_strategy()
print("REPLICAS: ", strategy.num_replicas_in_sync ) | Petals to the Metal - Flower Classification on TPU |
11,521,017 | X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.1,random_state=42 )<train_model> | AUTO = tf.data.experimental.AUTOTUNE
IMAGE_SIZE = [224, 224]
EPOCHS = 30
FOLDS = 3
SEED = 777
BATCH_SIZE = 16 * strategy.num_replicas_in_sync | Petals to the Metal - Flower Classification on TPU |
11,521,017 | vectorizer=TfidfVectorizer(ngram_range=(1,3),min_df=3,strip_accents='unicode',
use_idf=1,smooth_idf=1, sublinear_tf=1,max_features=None)
vectorizer.fit(list(df_train['text_processed'])+list(df_test['text_processed']))
print('vocab length',len(vectorizer.vocabulary_))<categorify> | MIXED_PRECISION = False
XLA_ACCELERATE = False
if MIXED_PRECISION:
if tpu: policy = tf.keras.mixed_precision.experimental.Policy('mixed_bfloat16')
else: policy = tf.keras.mixed_precision.experimental.Policy('mixed_float16')
mixed_precision.set_policy(policy)
print('Mixed precision enabled')
if XLA_ACCELERATE:
tf.config.optimizer.set_jit(True)
print('Accelerated Linear Algebra enabled' ) | Petals to the Metal - Flower Classification on TPU |
11,521,017 | X_train_onehot = vectorizer.transform(X_train ).todense()
X_val_onehot = vectorizer.transform(X_val ).todense()<train_model> | def decode_image(image_data):
image = tf.image.decode_jpeg(image_data, channels=3)
image = tf.cast(image, tf.float32)/ 255.0
image = tf.reshape(image, [*IMAGE_SIZE, 3])
return image
def read_labeled_tfrecord(example):
LABELED_TFREC_FORMAT = {
"image": tf.io.FixedLenFeature([], tf.string),
"class": tf.io.FixedLenFeature([], tf.int64),
}
example = tf.io.parse_single_example(example, LABELED_TFREC_FORMAT)
image = decode_image(example['image'])
label = tf.cast(example['class'], tf.int32)
return image, label
def read_unlabeled_tfrecord(example):
UNLABELED_TFREC_FORMAT = {
"image": tf.io.FixedLenFeature([], tf.string),
"id": tf.io.FixedLenFeature([], tf.string),
}
example = tf.io.parse_single_example(example, UNLABELED_TFREC_FORMAT)
image = decode_image(example['image'])
idnum = example['id']
return image, idnum
def load_dataset(filenames, labeled = True, ordered = False):
ignore_order = tf.data.Options()
if not ordered:
ignore_order.experimental_deterministic = False
dataset = tf.data.TFRecordDataset(filenames, num_parallel_reads = AUTO)
dataset = dataset.with_options(ignore_order)
dataset = dataset.map(read_labeled_tfrecord if labeled else read_unlabeled_tfrecord, num_parallel_calls = AUTO)
return dataset
def data_augment(image, label):
image = tf.image.random_flip_left_right(image)
return image, label
def get_training_dataset(dataset,do_aug=True):
dataset = dataset.map(data_augment, num_parallel_calls=AUTO)
if do_aug: dataset = dataset.map(transform, num_parallel_calls=AUTO)
dataset = dataset.repeat()
dataset = dataset.shuffle(2048)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.prefetch(AUTO)
return dataset
def get_validation_dataset(dataset):
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.cache()
dataset = dataset.prefetch(AUTO)
return dataset
def get_test_dataset(ordered=False):
dataset = load_dataset(TEST_FILENAMES, labeled=False, ordered=ordered)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.prefetch(AUTO)
return dataset
def count_data_items(filenames):
n = [int(re.compile(r"-([0-9]*)\." ).search(filename ).group(1)) for filename in filenames]
return np.sum(n)
NUM_TRAINING_IMAGES = int(count_data_items(TRAINING_FILENAMES)*(FOLDS-1.) /FOLDS)
NUM_VALIDATION_IMAGES = int(count_data_items(TRAINING_FILENAMES)*(1./FOLDS))
NUM_TEST_IMAGES = count_data_items(TEST_FILENAMES)
STEPS_PER_EPOCH = NUM_TRAINING_IMAGES // BATCH_SIZE
print('Dataset: {} training images, {} validation images, {} unlabeled test images'.format(NUM_TRAINING_IMAGES, NUM_VALIDATION_IMAGES, NUM_TEST_IMAGES)) | Petals to the Metal - Flower Classification on TPU |
11,521,017 | lr_clf = LogisticRegression(max_iter=150,penalty='l2',solver='lbfgs',random_state=0)
lr_clf.fit(X_train_onehot, y_train)
lr_pred = lr_clf.predict(X_val_onehot)
print('accuracy score: ',accuracy_score(lr_pred,y_val))
print(classification_report(y_val, lr_pred))<compute_test_metric> | def get_mat(rotation, shear, height_zoom, width_zoom, height_shift, width_shift):
rotation = math.pi * rotation / 180.
shear = math.pi * shear / 180.
c1 = tf.math.cos(rotation)
s1 = tf.math.sin(rotation)
one = tf.constant([1],dtype='float32')
zero = tf.constant([0],dtype='float32')
rotation_matrix = tf.reshape(tf.concat([c1,s1,zero, -s1,c1,zero, zero,zero,one],axis=0),[3,3])
c2 = tf.math.cos(shear)
s2 = tf.math.sin(shear)
shear_matrix = tf.reshape(tf.concat([one,s2,zero, zero,c2,zero, zero,zero,one],axis=0),[3,3])
zoom_matrix = tf.reshape(tf.concat([one/height_zoom,zero,zero, zero,one/width_zoom,zero, zero,zero,one],axis=0),[3,3])
shift_matrix = tf.reshape(tf.concat([one,zero,height_shift, zero,one,width_shift, zero,zero,one],axis=0),[3,3])
return K.dot(K.dot(rotation_matrix, shear_matrix), K.dot(zoom_matrix, shift_matrix))
def transform(image,label):
DIM = IMAGE_SIZE[0]
XDIM = DIM%2
rot = 15.* tf.random.normal([1],dtype='float32')
shr = 5.* tf.random.normal([1],dtype='float32')
h_zoom = 1.0 + tf.random.normal([1],dtype='float32')/10.
w_zoom = 1.0 + tf.random.normal([1],dtype='float32')/10.
h_shift = 16.* tf.random.normal([1],dtype='float32')
w_shift = 16.* tf.random.normal([1],dtype='float32')
m = get_mat(rot,shr,h_zoom,w_zoom,h_shift,w_shift)
x = tf.repeat(tf.range(DIM//2,-DIM//2,-1), DIM)
y = tf.tile(tf.range(-DIM//2,DIM//2),[DIM])
z = tf.ones([DIM*DIM],dtype='int32')
idx = tf.stack([x,y,z])
idx2 = K.dot(m,tf.cast(idx,dtype='float32'))
idx2 = K.cast(idx2,dtype='int32')
idx2 = K.clip(idx2,-DIM//2+XDIM+1,DIM//2)
idx3 = tf.stack([DIM//2-idx2[0,], DIM//2-1+idx2[1,]])
d = tf.gather_nd(image,tf.transpose(idx3))
return tf.reshape(d,[DIM,DIM,3]),label
| Petals to the Metal - Flower Classification on TPU |
11,521,017 | logloss_lr = log_loss(y_val,lr_clf.predict_proba(X_val_onehot))
print('logloss_lr:',logloss_lr )<compute_train_metric> | train_dataset_all = load_dataset(list(pd.DataFrame({'TRAINING_FILENAMES': TRAINING_FILENAMES})['TRAINING_FILENAMES']), labeled = True)
test_dataset_all = load_dataset(list(pd.DataFrame({'TEST_FILENAMES': TEST_FILENAMES})['TEST_FILENAMES']), labeled = True, ordered=True)
train_dataset = load_dataset(list(pd.DataFrame({'TRAINING_FILENAMES_ONLY': TRAINING_FILENAMES_ONLY})['TRAINING_FILENAMES_ONLY']), labeled = True)
val_dataset = load_dataset(list(pd.DataFrame({'VAL_FILENAMES_ONLY': VAL_FILENAMES_ONLY})['VAL_FILENAMES_ONLY']), labeled=True, ordered=True)
train_all = get_training_dataset(train_dataset_all)
test = get_test_dataset(test_dataset_all)
test_images_ds = test.map(lambda image, idnum: image)
train = get_training_dataset(train_dataset)
val = get_validation_dataset(val_dataset)
def Xception_model() :
with strategy.scope() :
rnet = Xception(
input_shape=(IMAGE_SIZE[0], IMAGE_SIZE[1], 3),
weights='imagenet',
include_top=False
)
rnet.trainable = True
model = tf.keras.Sequential([
rnet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation='softmax',dtype='float32')
])
model.compile(
optimizer='adam',
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
return model
def VGG16_model() :
with strategy.scope() :
rnet = VGG16(
input_shape=(IMAGE_SIZE[0], IMAGE_SIZE[1], 3),
weights='imagenet',
include_top=False
)
rnet.trainable = True
model = tf.keras.Sequential([
rnet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation='softmax',dtype='float32')
])
model.compile(
optimizer='adam',
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
return model
def DenseNet201_model() :
with strategy.scope() :
rnet = DenseNet201(
input_shape=(IMAGE_SIZE[0], IMAGE_SIZE[1], 3),
weights='imagenet',
include_top=False
)
rnet.trainable = True
model = tf.keras.Sequential([
rnet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation='softmax',dtype='float32')
])
model.compile(
optimizer='adam',
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
return model
def InceptionV3_model() :
with strategy.scope() :
rnet = InceptionV3(
input_shape=(IMAGE_SIZE[0], IMAGE_SIZE[1], 3),
weights='imagenet',
include_top=False
)
rnet.trainable = True
model = tf.keras.Sequential([
rnet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dropout(0.2),
tf.keras.layers.Dense(len(CLASSES), activation='softmax',dtype='float32')
])
model.compile(
optimizer='adam',
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
return model
def EfficientNetB7_model() :
with strategy.scope() :
rnet = EfficientNetB7(
input_shape=(IMAGE_SIZE[0], IMAGE_SIZE[1], 3),
weights='imagenet',
include_top=False
)
rnet.trainable = True
model = tf.keras.Sequential([
rnet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dropout(0.2),
tf.keras.layers.Dense(len(CLASSES), activation='softmax',dtype='float32')
])
model.compile(
optimizer='adam',
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
return model
def ResNet152V2_model() :
with strategy.scope() :
rnet = ResNet152V2(
input_shape=(IMAGE_SIZE[0], IMAGE_SIZE[1], 3),
weights='imagenet',
include_top=False
)
rnet.trainable = True
model = tf.keras.Sequential([
rnet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dropout(0.2),
tf.keras.layers.Dense(len(CLASSES), activation='softmax',dtype='float32')
])
model.compile(
optimizer='adam',
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
return model
def MobileNetV2_model() :
with strategy.scope() :
rnet = MobileNetV2(
input_shape=(IMAGE_SIZE[0], IMAGE_SIZE[1], 3),
weights='imagenet',
include_top=False
)
rnet.trainable = True
model = tf.keras.Sequential([
rnet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dropout(0.2),
tf.keras.layers.Dense(len(CLASSES), activation='softmax',dtype='float32')
])
model.compile(
optimizer='adam',
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
return model
def InceptionResNetV2_model() :
with strategy.scope() :
rnet = InceptionResNetV2(
input_shape=(IMAGE_SIZE[0], IMAGE_SIZE[1], 3),
weights='imagenet',
include_top=False
)
rnet.trainable = True
model = tf.keras.Sequential([
rnet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dropout(0.2),
tf.keras.layers.Dense(len(CLASSES), activation='softmax',dtype='float32')
])
model.compile(
optimizer='adam',
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
return model
models = {'Xception' : Xception_model,
'VGG16' : VGG16_model,
'DenseNet201' : DenseNet201_model,
'InceptionV3' : InceptionV3_model,
'InceptionResNetV2' : InceptionResNetV2_model
}
historys = {}
predictions = {}
predictions_val = {}
predictions_prob = {}
times = {}
MODELS_NUMBER = 5
| Petals to the Metal - Flower Classification on TPU |
11,521,017 | mnb_clf = MultinomialNB()
mnb_clf.fit(X_train_onehot, y_train)
mnb_pred = mnb_clf.predict(X_val_onehot)
print('accuracy score: ',accuracy_score(mnb_pred,y_val))
print(classification_report(y_val, mnb_pred))<compute_test_metric> | df = pd.DataFrame(predictions)
pred = []
for i in range(0, 7382):
if df.loc[i,:].unique().shape[0] < MODELS_NUMBER :
pred.append(stats.mode(df.loc[i,:].values)[0][0])
else :
pred.append(df.loc[i,'InceptionResNetV2'])
df = pd.DataFrame(predictions_val)
pred_val = []
for i in range(0, 3712):
if df.loc[i,:].unique().shape[0] < MODELS_NUMBER :
pred_val.append(stats.mode(df.loc[i,:].values)[0][0])
else :
pred_val.append(df.loc[i,'InceptionResNetV2'])
avg_prob = predictions_prob['Xception'] + predictions_prob['VGG16'] + predictions_prob['DenseNet201'] + predictions_prob['InceptionV3'] + predictions_prob['InceptionResNetV2']
pred_avg = pd.DataFrame(np.argmax(avg_prob, axis=-1)) | Petals to the Metal - Flower Classification on TPU |
11,521,017 | <predict_on_test><EOS> | test_ids_ds = test.map(lambda image, idnum: idnum ).unbatch()
test_ids = next(iter(test_ids_ds.batch(NUM_TEST_IMAGES)) ).numpy().astype('U')
np.savetxt('submission_vote.csv', np.rec.fromarrays([test_ids, pred]), fmt=['%s', '%d'], delimiter=',', header='id,label', comments='')
np.savetxt('submission.csv', np.rec.fromarrays([test_ids, np.argmax(avg_prob, axis=-1)]), fmt=['%s', '%d'], delimiter=',', header='id,label', comments='' ) | Petals to the Metal - Flower Classification on TPU |
11,593,914 | <SOS> metric: macrofscore Kaggle data source: petals-to-the-metal-flower-classification-on-tpu<compute_test_metric> | !pip install -q efficientnet | Petals to the Metal - Flower Classification on TPU |
11,593,914 | logloss_rf = log_loss(y_val,rf_clf.predict_proba(X_val_onehot))
print('logloss_rf:',logloss_rf )<train_model> | import numpy as np
import pandas as pd
import re
import os
import efficientnet.tfkeras as efn
import tensorflow as tf
from tensorflow.keras.layers import Input, Dense
from tensorflow.keras.layers import GlobalAveragePooling2D
from tensorflow.keras.models import Sequential
from tensorflow.keras.callbacks import LearningRateScheduler
from kaggle_datasets import KaggleDatasets
from sklearn.metrics import confusion_matrix, f1_score
from sklearn.metrics import precision_score, recall_score
import seaborn as sea
import matplotlib.pyplot as plt | Petals to the Metal - Flower Classification on TPU |
11,593,914 | xgb_clf = xgb.XGBClassifier(n_estimators=100,n_jobs=-1,max_depth=15,
min_child_weight=3,objective='binary:logistic'
,colsample_bytree=0.4)
xgb_clf.fit(X_train_onehot, y_train)
xgb_predictions = xgb_clf.predict(X_val_onehot)
print('accuracy score: ',accuracy_score(xgb_predictions,y_val))
print(classification_report(y_val, xgb_predictions))<compute_test_metric> | sea.set_style("darkgrid")
np.random.seed(3)
tf.random.set_seed(6 ) | Petals to the Metal - Flower Classification on TPU |
11,593,914 | logloss_xgb = log_loss(y_val,xgb_clf.predict_proba(X_val_onehot))
print('logloss_rf:',logloss_xgb )<choose_model_class> | gcs_path = KaggleDatasets().get_gcs_path("tpu-getting-started")
data_path = os.path.join(gcs_path, "tfrecords-jpeg-512x512/")
train_path = os.path.join(gcs_path, "tfrecords-jpeg-512x512/train/")
val_path = os.path.join(gcs_path, "tfrecords-jpeg-512x512/val/")
test_path = os.path.join(gcs_path, "tfrecords-jpeg-512x512/test/" ) | Petals to the Metal - Flower Classification on TPU |
11,593,914 | model = Sequential()
model.add(Dense(512, activation='relu', input_dim=np.shape(X_train_onehot)[1],
kernel_regularizer=regularizers.l2(0.01),
activity_regularizer=regularizers.l1(0.01)))
model.add(BatchNormalization())
model.add(Dropout(0.5))
model.add(Dense(256, activation='relu'))
model.add(BatchNormalization())
model.add(Dropout(0.6))
model.add(Dense(64, activation='relu'))
model.add(BatchNormalization())
model.add(Dropout(0.5))
model.add(Dense(1, activation='sigmoid'))
adam = Adam(lr=1e-4, beta_1=0.9, beta_2=0.999, epsilon=10**-8, decay=0.0001, amsgrad=False)
model.compile(optimizer= adam,
loss='binary_crossentropy',
metrics=['accuracy'])
print(model.summary())
hist = model.fit(X_train_onehot, y_train,validation_data =(X_val_onehot,y_val), epochs=20, batch_size=16 )<predict_on_test> | classes = ['pink primrose', 'hard-leaved pocket orchid',
'canterbury bells', 'sweet pea',
'wild geranium', 'tiger lily',
'moon orchid', 'bird of paradise',
'monkshood', 'globe thistle',
'snapdragon', "colt's foot",
'king protea', 'spear thistle',
'yellow iris', 'globe-flower',
'purple coneflower', 'peruvian lily',
'balloon flower', 'giant white arum lily',
'fire lily', 'pincushion flower',
'fritillary', 'red ginger',
'grape hyacinth', 'corn poppy',
'prince of wales feathers', 'stemless gentian',
'artichoke', 'sweet william',
'carnation', 'garden phlox',
'love in the mist', 'cosmos',
'alpine sea holly', 'ruby-lipped cattleya',
'cape flower', 'great masterwort',
'siam tulip', 'lenten rose',
'barberton daisy', 'daffodil',
'sword lily', 'poinsettia',
'bolero deep blue', 'wallflower',
'marigold', 'buttercup',
'daisy', 'common dandelion',
'petunia', 'wild pansy',
'primula', 'sunflower',
'lilac hibiscus', 'bishop of llandaff',
'gaura', 'geranium',
'orange dahlia', 'pink-yellow dahlia',
'cautleya spicata', 'japanese anemone',
'black-eyed susan', 'silverbush',
'californian poppy', 'osteospermum',
'spring crocus', 'iris',
'windflower', 'tree poppy',
'gazania', 'azalea',
'water lily', 'rose',
'thorn apple', 'morning glory',
'passion flower', 'lotus',
'toad lily', 'anthurium',
'frangipani', 'clematis',
'hibiscus', 'columbine',
'desert-rose', 'tree mallow',
'magnolia', 'cyclamen ',
'watercress', 'canna lily',
'hippeastrum ', 'bee balm',
'pink quill', 'foxglove',
'bougainvillea', 'camellia',
'mallow', 'mexican petunia',
'bromelia', 'blanket flower',
'trumpet creeper', 'blackberry lily',
'common tulip', 'wild rose'] | Petals to the Metal - Flower Classification on TPU |
11,593,914 | dnn_pred = model.predict_classes(X_val_onehot )<compute_test_metric> | def read_record(example):
features = {
"image": tf.io.FixedLenFeature([], tf.string),
"class": tf.io.FixedLenFeature([], tf.int64),
}
example = tf.io.parse_single_example(example, features)
image = tf.image.decode_jpeg(example["image"], channels=3)
image = tf.cast(image, tf.float32)/ 255.0
image = tf.reshape(image, [512,512,3])
label = tf.cast(example["class"], tf.int32)
return image, label
def read_record_test(example):
features = {
"image": tf.io.FixedLenFeature([], tf.string),
"id": tf.io.FixedLenFeature([], tf.string),
}
example = tf.io.parse_single_example(example, features)
image = tf.image.decode_jpeg(example["image"], channels=3)
image = tf.cast(image, tf.float32)/ 255.0
image = tf.reshape(image, [512,512,3])
image_id = example["id"]
return image, image_id | Petals to the Metal - Flower Classification on TPU |
11,593,914 | print('accuracy score: ',accuracy_score(dnn_pred,y_val))
print(classification_report(y_val, dnn_pred))<compute_test_metric> | AUTO = tf.data.experimental.AUTOTUNE
def get_size(filenames):
data_size = 0
for i in range(len(filenames)) :
count = re.search(r"-[0-9][0-9][0-9]\.", filenames[i] ).group()
data_size = data_size + int(count[1:4])
return data_size
def prepare_dataset(flag, order = False):
filenames = tf.io.gfile.glob(os.path.join(data_path, flag)+"/*.tfrec")
data_size = get_size(filenames)
dataset = tf.data.TFRecordDataset(filenames, num_parallel_reads=AUTO)
ignore_order = tf.data.Options()
if order == False:
ignore_order.experimental_deterministic = False
else:
ignore_order.experimental_deterministic = True
dataset = dataset.with_options(ignore_order)
if flag == "test":
dataset = dataset.map(read_record_test, num_parallel_calls=AUTO)
else:
dataset = dataset.map(read_record, num_parallel_calls=AUTO)
return dataset, data_size
train_dataset, train_size = prepare_dataset("train")
val_dataset, val_size = prepare_dataset("val")
test_dataset, test_size = prepare_dataset("test")
print("Size of Training Dataset\t: ", train_size)
print("Size of Validation Dataset\t: ", val_size)
print("Size of Test Dataset\t\t: ", test_size ) | Petals to the Metal - Flower Classification on TPU |
11,593,914 | logloss_dnn = log_loss(y_val,model.predict_proba(X_val_onehot))
print('logloss_dnn:',logloss_dnn )<predict_on_test> | def augment(image, label):
cont_low = 0.8
cont_high = 1.2
cont_factor = tf.random.uniform([], minval=cont_low,
maxval=cont_high,
dtype=tf.float32)
trn_image = tf.image.adjust_contrast(image, cont_factor)
return trn_image, label | Petals to the Metal - Flower Classification on TPU |
11,593,914 | lr_predictions_val = lr_clf.predict_proba(X_val_onehot)
mnb_predictions_val = mnb_clf.predict_proba(X_val_onehot)
xgb_predictions_val = xgb_clf.predict_proba(X_val_onehot)
<define_variables> | batch_size = 16 * tpu_strategy.num_replicas_in_sync
train_dataset = train_dataset.map(augment,
num_parallel_calls = AUTO)
train_dataset = train_dataset.repeat().shuffle(3000)\
.batch(batch_size ).prefetch(AUTO)
val_dataset = val_dataset.batch(batch_size ).prefetch(AUTO)
test_dataset = test_dataset.batch(batch_size ).prefetch(AUTO ) | Petals to the Metal - Flower Classification on TPU |
11,593,914 | predictions_val = 1/3*lr_predictions_val[:,1]+1/3*mnb_predictions_val[:,1] \
+1/3*xgb_predictions_val[:,1]
predictions_val = np.where(predictions_val>0.5, 1, 0 )<compute_test_metric> | with tpu_strategy.scope() :
base_model = efn.EfficientNetB5(include_top=False,
input_shape=(512,512,3),
weights='imagenet')
base_model.trainable = True
model = Sequential(name="Flower_Detector")
model.add(base_model)
model.add(GlobalAveragePooling2D(name="GAP"))
model.add(Dense(104, activation="softmax", name="Probs"))
model.compile(optimizer=tf.keras.optimizers.Adam() ,
loss="sparse_categorical_crossentropy",
metrics=["sparse_categorical_accuracy"])
model.summary() | Petals to the Metal - Flower Classification on TPU |
11,593,914 | print('accuracy score: ',accuracy_score(predictions_val,y_val))
print(classification_report(y_val, predictions_val))<load_from_csv> | step_per_epoch = train_size // batch_size
history = model.fit(train_dataset,
validation_data=val_dataset,
steps_per_epoch=step_per_epoch,
epochs=epoch,
callbacks=[lrs] ) | Petals to the Metal - Flower Classification on TPU |
11,593,914 | df_test = pd.read_csv('.. /input/nlp-getting-started/test.csv' )<feature_engineering> | val_dataset, val_size = prepare_dataset("val", True)
val_dataset = val_dataset.batch(batch_size ).prefetch(AUTO)
val_image_dataset = val_dataset.map(lambda image,label: image)
val_label_dataset = val_dataset.map(lambda image,label: label ).unbatch()
val_labels = next(iter(val_label_dataset.batch(val_size)) ).numpy()
val_probs = model.predict(val_image_dataset)
val_preds = np.argmax(val_probs, axis=-1 ) | Petals to the Metal - Flower Classification on TPU |
11,593,914 | df_test['text_processed'] = df_test['text'].apply(text_preprocessing )<categorify> | con_mat = confusion_matrix(val_labels, val_preds,
labels=range(len(classes)))
prec = precision_score(val_labels, val_preds,
labels=range(len(classes)) ,
average='macro')
rec = recall_score(val_labels, val_preds,
labels=range(len(classes)) ,
average='macro')
f1 = f1_score(val_labels, val_preds,
labels=range(len(classes)) ,
average='macro')
plt.figure(figsize=(10,10))
ax = plt.gca()
ax.matshow(con_mat, cmap='Greens')
ax.set_xticks(range(len(classes)))
ax.set_xticklabels(classes, fontdict={'fontsize': 6})
plt.setp(ax.get_xticklabels() , rotation=65, ha="left",
rotation_mode="anchor")
ax.set_yticks(range(len(classes)))
ax.set_yticklabels(classes, fontdict={'fontsize': 6})
plt.setp(ax.get_yticklabels() , rotation=25, ha="right",
rotation_mode="anchor")
plt.show()
print('Precision \t: {:.4f}'.format(prec))
print('Recall \t\t: {:.4f}'.format(rec))
print('F1 score \t: {:.4f}'.format(f1)) | Petals to the Metal - Flower Classification on TPU |
11,593,914 | X_test = df_test['text_processed']
X_test_onehot = vectorizer.transform(X_test ).todense()<predict_on_test> | test_dataset, test_size = prepare_dataset("test", True)
test_dataset = test_dataset.batch(batch_size ).prefetch(AUTO)
test_image_dataset = test_dataset.map(lambda image,idnum: image)
test_id_dataset = test_dataset.map(lambda image,idnum: idnum ).unbatch()
test_ids = next(iter(test_id_dataset.batch(test_size)) ).numpy().astype('U')
test_probs = model.predict(test_image_dataset)
test_preds = np.argmax(test_probs, axis=-1 ) | Petals to the Metal - Flower Classification on TPU |
11,414,583 | lr_predictions = lr_clf.predict_proba(X_test_onehot)
mnb_predictions = mnb_clf.predict_proba(X_test_onehot)
xgb_predictions = xgb_clf.predict_proba(X_test_onehot)
<define_variables> | !pip install -q efficientnet | Petals to the Metal - Flower Classification on TPU |
11,414,583 | predictions = 1/3*lr_predictions[:,1]+1/3*mnb_predictions[:,1]+1/3*xgb_predictions[:,1]<prepare_output> | print("Tensorflow version " + tf.__version__ ) | Petals to the Metal - Flower Classification on TPU |
11,414,583 | predictions = np.where(predictions>0.5, 1, 0 )<save_to_csv> | AUTO = tf.data.experimental.AUTOTUNE
tpu = tf.distribute.cluster_resolver.TPUClusterResolver()
tf.config.experimental_connect_to_cluster(tpu)
tf.tpu.experimental.initialize_tpu_system(tpu)
strategy = tf.distribute.experimental.TPUStrategy(tpu)
GCS_DS_PATH = KaggleDatasets().get_gcs_path()
IMAGE_SIZE = [512, 512]
EPOCHS = 50
BATCH_SIZE = 16 * strategy.num_replicas_in_sync
GCS_PATH_SELECT = {
192: GCS_DS_PATH + '/tfrecords-jpeg-192x192',
224: GCS_DS_PATH + '/tfrecords-jpeg-224x224',
331: GCS_DS_PATH + '/tfrecords-jpeg-331x331',
512: GCS_DS_PATH + '/tfrecords-jpeg-512x512'
}
GCS_PATH = GCS_PATH_SELECT[IMAGE_SIZE[0]]
TRAINING_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/train/*.tfrec')
VALIDATION_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/val/*.tfrec')
TEST_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/test/*.tfrec' ) | Petals to the Metal - Flower Classification on TPU |
11,414,583 | df_submission = pd.read_csv('.. /input/nlp-getting-started/sample_submission.csv')
df_submission['target'] = predictions
df_submission.to_csv('submission.csv',index=False )<set_options> | CLASSES = ['pink primrose', 'hard-leaved pocket orchid', 'canterbury bells', 'sweet pea', 'wild geranium', 'tiger lily', 'moon orchid', 'bird of paradise', 'monkshood', 'globe thistle',
'snapdragon', "colt's foot", 'king protea', 'spear thistle', 'yellow iris', 'globe-flower', 'purple coneflower', 'peruvian lily', 'balloon flower', 'giant white arum lily',
'fire lily', 'pincushion flower', 'fritillary', 'red ginger', 'grape hyacinth', 'corn poppy', 'prince of wales feathers', 'stemless gentian', 'artichoke', 'sweet william',
'carnation', 'garden phlox', 'love in the mist', 'cosmos', 'alpine sea holly', 'ruby-lipped cattleya', 'cape flower', 'great masterwort', 'siam tulip', 'lenten rose',
'barberton daisy', 'daffodil', 'sword lily', 'poinsettia', 'bolero deep blue', 'wallflower', 'marigold', 'buttercup', 'daisy', 'common dandelion',
'petunia', 'wild pansy', 'primula', 'sunflower', 'lilac hibiscus', 'bishop of llandaff', 'gaura', 'geranium', 'orange dahlia', 'pink-yellow dahlia',
'cautleya spicata', 'japanese anemone', 'black-eyed susan', 'silverbush', 'californian poppy', 'osteospermum', 'spring crocus', 'iris', 'windflower', 'tree poppy',
'gazania', 'azalea', 'water lily', 'rose', 'thorn apple', 'morning glory', 'passion flower', 'lotus', 'toad lily', 'anthurium',
'frangipani', 'clematis', 'hibiscus', 'columbine', 'desert-rose', 'tree mallow', 'magnolia', 'cyclamen ', 'watercress', 'canna lily',
'hippeastrum ', 'bee balm', 'pink quill', 'foxglove', 'bougainvillea', 'camellia', 'mallow', 'mexican petunia', 'bromelia', 'blanket flower',
'trumpet creeper', 'blackberry lily', 'common tulip', 'wild rose'] | Petals to the Metal - Flower Classification on TPU |
11,414,583 | pd.set_option('display.max_rows', 1000)
seed =45
plt.style.use('fivethirtyeight' )<load_from_csv> | def decode_image(image_data):
image = tf.image.decode_jpeg(image_data, channels=3)
image = tf.cast(image, tf.float32)/ 255.0
image = tf.reshape(image, [*IMAGE_SIZE, 3])
return image
def read_labeled_tfrecord(example):
LABELED_TFREC_FORMAT = {
"image": tf.io.FixedLenFeature([], tf.string),
"class": tf.io.FixedLenFeature([], tf.int64),
}
example = tf.io.parse_single_example(example, LABELED_TFREC_FORMAT)
image = decode_image(example['image'])
label = tf.cast(example['class'], tf.int32)
return image, label
def read_unlabeled_tfrecord(example):
UNLABELED_TFREC_FORMAT = {
"image": tf.io.FixedLenFeature([], tf.string),
"id": tf.io.FixedLenFeature([], tf.string),
}
example = tf.io.parse_single_example(example, UNLABELED_TFREC_FORMAT)
image = decode_image(example['image'])
idnum = example['id']
return image, idnum
def load_dataset(filenames, labeled=True, ordered=False):
ignore_order = tf.data.Options()
if not ordered:
ignore_order.experimental_deterministic = False
dataset = tf.data.TFRecordDataset(filenames, num_parallel_reads=AUTO)
dataset = dataset.with_options(ignore_order)
dataset = dataset.map(read_labeled_tfrecord if labeled else read_unlabeled_tfrecord, num_parallel_calls=AUTO)
return dataset
def data_augment(image, label):
image = tf.image.random_flip_left_right(image)
return image, label
def get_training_dataset() :
dataset = load_dataset(TRAINING_FILENAMES, labeled=True)
dataset = dataset.map(data_augment, num_parallel_calls=AUTO)
dataset = dataset.repeat()
dataset = dataset.shuffle(2048)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.prefetch(AUTO)
return dataset
def get_validation_dataset(ordered=False):
dataset = load_dataset(VALIDATION_FILENAMES, labeled=True, ordered=ordered)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.cache()
dataset = dataset.prefetch(AUTO)
return dataset
def get_test_dataset(ordered=False):
dataset = load_dataset(TEST_FILENAMES, labeled=False, ordered=ordered)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.prefetch(AUTO)
return dataset
def count_data_items(filenames):
n = [int(re.compile(r"-([0-9]*)\." ).search(filename ).group(1)) for filename in filenames]
return np.sum(n)
NUM_TRAINING_IMAGES = count_data_items(TRAINING_FILENAMES)
NUM_VALIDATION_IMAGES = count_data_items(VALIDATION_FILENAMES)
NUM_TEST_IMAGES = count_data_items(TEST_FILENAMES)
STEPS_PER_EPOCH = NUM_TRAINING_IMAGES // BATCH_SIZE
print('Dataset: {} training images, {} validation images, {} unlabeled test images'.format(NUM_TRAINING_IMAGES, NUM_VALIDATION_IMAGES, NUM_TEST_IMAGES)) | Petals to the Metal - Flower Classification on TPU |
11,414,583 | train = pd.read_csv('/kaggle/input/titanic/train.csv')
test = pd.read_csv('/kaggle/input/titanic/test.csv' )<feature_engineering> | with strategy.scope() :
enet = efn.EfficientNetB7(
input_shape=(812, 812, 3),
weights='imagenet',
include_top=False
)
model = tf.keras.Sequential([
enet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation='softmax')
])
model.compile(
optimizer=tf.keras.optimizers.Adam(lr=0.0001),
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
model.summary() | Petals to the Metal - Flower Classification on TPU |
11,414,583 | train.loc[train['PassengerId'] == 631, 'Age'] = 48
train.loc[train['PassengerId'] == 69, ['SibSp', 'Parch']] = [0,0]
test.loc[test['PassengerId'] == 1106, ['SibSp', 'Parch']] = [0,0]<sort_values> | scheduler = tf.keras.callbacks.ReduceLROnPlateau(patience=3, verbose=1)
history = model.fit(
get_training_dataset() ,
steps_per_epoch=STEPS_PER_EPOCH,
epochs=EPOCHS,
callbacks=[scheduler],
validation_data=get_validation_dataset()
) | Petals to the Metal - Flower Classification on TPU |
11,414,583 | <categorify><EOS> | test_ds = get_test_dataset(ordered=True)
print('Computing predictions...')
test_images_ds = test_ds.map(lambda image, idnum: image)
probabilities = model.predict(test_images_ds)
predictions = np.argmax(probabilities, axis=-1)
print(predictions)
print('Generating submission.csv file...')
test_ids_ds = test_ds.map(lambda image, idnum: idnum ).unbatch()
test_ids = next(iter(test_ids_ds.batch(NUM_TEST_IMAGES)) ).numpy().astype('U')
np.savetxt('submission.csv', np.rec.fromarrays([test_ids, predictions]), fmt=['%s', '%d'], delimiter=',', header='id,label', comments='' ) | Petals to the Metal - Flower Classification on TPU |
11,225,265 | <SOS> metric: macrofscore Kaggle data source: petals-to-the-metal-flower-classification-on-tpu<drop_column> | !pip install efficientnet --quiet
sns.set() | Petals to the Metal - Flower Classification on TPU |
11,225,265 |
<concatenate> | flower_categories = [
'pink primrose', 'hard-leaved pocket orchid', 'canterbury bells',
'sweet pea', 'wild geranium', 'tiger lily',
'moon orchid', 'bird of paradise', 'monkshood',
'globe thistle', 'snapdragon', 'colt's foot',
'king protea', 'spear thistle', 'yellow iris',
'globe-flower', 'purple coneflower', 'peruvian lily',
'balloon flower', 'giant white arum lily', 'fire lily',
'pincushion flower', 'fritillary', 'red ginger',
'grape hyacinth', 'corn poppy', 'prince of wales feathers',
'stemless gentian', 'artichoke', 'sweet william',
'carnation', 'garden phlox', 'love in the mist',
'cosmos', 'alpine sea holly', 'ruby-lipped cattleya',
'cape flower', 'great masterwort', 'siam tulip',
'lenten rose', 'barberton daisy', 'daffodil',
'sword lily', 'poinsettia', 'bolero deep blue',
'wallflower', 'marigold', 'buttercup',
'daisy', 'common dandelion', 'petunia',
'wild pansy', 'primula', 'sunflower',
'lilac hibiscus', 'bishop of llandaff', 'gaura',
'geranium', 'orange dahlia', 'pink-yellow dahlia',
'cautleya spicata', 'japanese anemone', 'black-eyed susan',
'silverbush', 'californian poppy', 'osteospermum',
'spring crocus', 'iris', 'windflower',
'tree poppy', 'gazania', 'azalea',
'water lily', 'rose', 'thorn apple',
'morning glory', 'passion flower', 'lotus',
'toad lily', 'anthurium', 'frangipani',
'clematis', 'hibiscus', 'columbine',
'desert-rose', 'tree mallow', 'magnolia',
'cyclamen ', 'watercress', 'canna lily',
'hippeastrum ', 'bee balm', 'pink quill',
'foxglove', 'bougainvillea', 'camellia',
'mallow', 'mexican petunia', 'bromelia',
'blanket flower', 'trumpet creeper', 'blackberry lily',
'common tulip', 'wild rose'
]
print('Number of flower categories:', len(flower_categories))
| Petals to the Metal - Flower Classification on TPU |
11,225,265 | df = pd.concat(( train.loc[:,'Pclass':'Embarked'], test.loc[:,'Pclass':'Embarked'])).reset_index(drop=True)
<load_from_csv> | image_size =(224,224)
data_gcs = KaggleDatasets().get_gcs_path('tpu-getting-started')
ext_gcs = KaggleDatasets().get_gcs_path('tf-flower-photo-tfrec')
data_dir_by_size = {
(512, 512): '/tfrecords-jpeg-512x512',
(331, 331): '/tfrecords-jpeg-331x331',
(224, 224): '/tfrecords-jpeg-224x224',
(192, 192): '/tfrecords-jpeg-192x192'
}
subdir = data_dir_by_size[image_size]
train_file_names = tf.io.gfile.glob(data_gcs + subdir+ '/train'+ '/*tfrec')
val_file_names = tf.io.gfile.glob(data_gcs + subdir + '/val' + '/*tfrec')
test_file_names = tf.io.gfile.glob(data_gcs + subdir + '/test' + '/*tfrec')
imagenet_files = tf.io.gfile.glob(ext_gcs + '/imagenet' + subdir + '/*.tfrec')
inaturelist_files = tf.io.gfile.glob(ext_gcs + '/inaturalist' + subdir + '/*.tfrec')
openimage_files = tf.io.gfile.glob(ext_gcs + '/openimage' + subdir + '/*.tfrec')
oxford_files = tf.io.gfile.glob(ext_gcs + '/oxford_102' + subdir + '/*.tfrec')
tensorflow_files = tf.io.gfile.glob(ext_gcs + '/tf_flowers' + subdir + '/*.tfrec')
train_file_names = train_file_names + imagenet_files + inaturelist_files + \
openimage_files + oxford_files + tensorflow_files | Petals to the Metal - Flower Classification on TPU |
11,225,265 | %pip install autoviz
AV = AutoViz_Class()
report_2 = AV.AutoViz("/kaggle/input/titanic/train.csv" )<feature_engineering> | def decode_image(image_data):
image = tf.image.decode_jpeg(image_data, channels =3)
image = tf.cast(image, tf.float32)/ 255.0
image = tf.reshape(image, [*image_size,3])
return image
def read_labeled_tfrecord(example):
example = tf.io.parse_single_example(
serialized = example,
features = {
'image' : tf.io.FixedLenFeature([], tf.string),
'class' : tf.io.FixedLenFeature([], tf.int64)
}
)
image = decode_image(example['image'])
label = tf.cast(example['class'], tf.int32)
return image, label
def read_unlabeled_tfrecord(example):
example = tf.io.parse_single_example(
serialized=example,
features={
'image': tf.io.FixedLenFeature([], tf.string),
'id': tf.io.FixedLenFeature([], tf.string),
}
)
image = decode_image(example['image'])
idnum = example['id']
return image, idnum
def load_dataset(filenames, labeled = True, ordered= False):
ignore_order = tf.data.Options()
ignore_order.experimental_deterministic = ordered
dataset = tf.data.TFRecordDataset(
filenames, num_parallel_reads = tf.data.experimental.AUTOTUNE)
dataset = dataset.with_options(ignore_order)
dataset = dataset.map(
read_labeled_tfrecord if labeled else read_unlabeled_tfrecord,
num_parallel_calls=tf.data.experimental.AUTOTUNE)
return dataset
def count_data_items(filenames):
return np.sum([
int(re.compile(r"-([0-9]*)\." ).search(filename ).group(1))
for filename in filenames])
train_dataset = load_dataset(train_file_names, labeled=True, ordered=False)
val_dataset = load_dataset(val_file_names, labeled=True, ordered=False)
test_dataset = load_dataset(test_file_names, labeled=False, ordered=True)
num_training_samples = count_data_items(train_file_names)
num_validation_samples = count_data_items(val_file_names)
num_testing_samples = count_data_items(test_file_names)
| Petals to the Metal - Flower Classification on TPU |
11,225,265 | def basic_details(df):
b = pd.DataFrame()
b['Missing value, %'] = round(df.isnull().sum() /df.shape[0]*100)
b['N unique value'] = df.nunique()
b['dtype'] = df.dtypes
return b
basic_details(df )<categorify> | print('Train samples:', num_training_samples, end=', ')
print('Val samples:', num_validation_samples, end=', ')
print('Test samples:', num_testing_samples ) | Petals to the Metal - Flower Classification on TPU |
11,225,265 | df['Title'] = df.Name.str.extract('([A-Za-z]+)\.', expand=False)
df['Title'] = df['Title'].replace(['Lady', 'Countess','Capt', 'Col', 'Don', 'Dr', 'Major', 'Rev', 'Sir', 'Jonkheer', 'Dona'], 'Rare')
df['Title'] = df['Title'].replace('Mlle', 'Miss')
df['Title'] = df['Title'].replace('Ms', 'Miss')
df['Title'] = df['Title'].replace('Mme', 'Mrs')
title_mapping = {"Mr": 1, "Miss": 2, "Mrs": 3, "Master": 4, "Rare": 5}
df['Title'] = df['Title'].map(title_mapping)
df['Title'] = df['Title'].fillna(0)
df = df.drop(['Name'], axis=1)
df["Sex"][df["Sex"] == "male"] = 0
df["Sex"][df["Sex"] == "female"] = 1
df["Sex"] = df["Sex"].astype(int)
df['Age'] = df.groupby('Pclass')['Age'].transform(lambda x: x.fillna(x.median()))
df["Age"] = df["Age"].astype(int)
df['Age_cat'] = pd.qcut(df['Age'],q=[0,.16,.33,.49,.66,.83, 1], labels=False, precision=1)
tickets = []
for i in list(df.Ticket):
if not i.isdigit() :
tickets.append(i.replace(".","" ).replace("/","" ).strip().split(" ")[0])
else:
tickets.append("x")
df["Ticket"] = tickets
df = pd.get_dummies(df, columns= ["Ticket"], prefix = "T")
df['Fare'] = df.groupby("Pclass")['Fare'].transform(lambda x: x.fillna(x.median()))
df['Zero_Fare'] = df['Fare'].map(lambda x: 1 if x == 0 else(0))
def fare_category(fr):
if fr <= 7.91:
return 1
elif fr <= 14.454 and fr > 7.91:
return 2
elif fr <= 31 and fr > 14.454:
return 3
return 4
df['Fare_cat'] = df['Fare'].apply(fare_category)
df['Cabin'] = df['Cabin'].fillna('U')
df['Cabin'] = df['Cabin'].map(lambda x: re.compile("([a-zA-Z]+)" ).search(x ).group())
cabin_category = {'A':9, 'B':8, 'C':7, 'D':6, 'E':5, 'F':4, 'G':3, 'T':2, 'U':1}
df['Cabin'] = df['Cabin'].map(cabin_category)
df["Embarked"] = df["Embarked"].fillna("C")
df["Embarked"][df["Embarked"] == "S"] = 1
df["Embarked"][df["Embarked"] == "C"] = 2
df["Embarked"][df["Embarked"] == "Q"] = 3
df["Embarked"] = df["Embarked"].astype(int)
df['FamilySize'] = df['SibSp'] + df['Parch'] + 1
df['FamilySize_cat'] = df['FamilySize'].map(lambda x: 1 if x == 1
else(2 if 5 > x >= 2
else(3 if 8 > x >= 5
else 4)
))
df['Alone'] = [1 if i == 1 else 0 for i in df['FamilySize']]
dummy_col=['Title', 'Sex', 'Age_cat', 'SibSp', 'Parch', 'Fare_cat', 'Cabin', 'Embarked', 'Pclass', 'FamilySize_cat']
dummy = pd.get_dummies(df[dummy_col], columns=dummy_col, drop_first=False)
df = pd.concat([dummy, df], axis = 1)
df['FareCat_Sex'] = df['Fare_cat']*df['Sex']
df['Pcl_Sex'] = df['Pclass']*df['Sex']
df['Pcl_Title'] = df['Pclass']*df['Title']
df['Age_cat_Sex'] = df['Age_cat']*df['Sex']
df['Age_cat_Pclass'] = df['Age_cat']*df['Pclass']
df['Title_Sex'] = df['Title']*df['Sex']
df['Age_Fare'] = df['Age_cat']*df['Fare_cat']
df['SmallF'] = df['FamilySize'].map(lambda s: 1 if s == 2 else 0)
df['MedF'] = df['FamilySize'].map(lambda s: 1 if 3 <= s <= 4 else 0)
df['LargeF'] = df['FamilySize'].map(lambda s: 1 if s >= 5 else 0)
df['Senior'] = df['Age'].map(lambda s:1 if s>70 else 0)
<feature_engineering> | def get_mat(rotation, shear, height_zoom, width_zoom, height_shift, width_shift):
rotation = math.pi * rotation / 180.
shear = math.pi * shear / 180.
c1 = tf.math.cos(rotation)
s1 = tf.math.sin(rotation)
one = tf.constant([1],dtype='float32')
zero = tf.constant([0],dtype='float32')
rotation_matrix = tf.reshape(tf.concat([c1,s1,zero, -s1,c1,zero, zero,zero,one],axis=0),[3,3])
c2 = tf.math.cos(shear)
s2 = tf.math.sin(shear)
shear_matrix = tf.reshape(tf.concat([one,s2,zero, zero,c2,zero, zero,zero,one],axis=0),[3,3])
zoom_matrix = tf.reshape(tf.concat([one/height_zoom,zero,zero, zero,one/width_zoom,zero, zero,zero,one],axis=0),[3,3])
shift_matrix = tf.reshape(tf.concat([one,zero,height_shift, zero,one,width_shift, zero,zero,one],axis=0),[3,3])
return K.dot(K.dot(rotation_matrix, shear_matrix), K.dot(zoom_matrix, shift_matrix)) | Petals to the Metal - Flower Classification on TPU |
11,225,265 | def descrictive_stat_feat(df):
df = pd.DataFrame(df)
dcol= [c for c in df.columns if df[c].nunique() >=10]
d_median = df[dcol].median(axis=0)
d_mean = df[dcol].mean(axis=0)
q1 = df[dcol].apply(np.float32 ).quantile(0.25)
q3 = df[dcol].apply(np.float32 ).quantile(0.75)
for c in dcol:
df[c+str('_median_range')] =(df[c].astype(np.float32 ).values > d_median[c] ).astype(np.int8)
df[c+str('_mean_range')] =(df[c].astype(np.float32 ).values > d_mean[c] ).astype(np.int8)
df[c+str('_q1')] =(df[c].astype(np.float32 ).values < q1[c] ).astype(np.int8)
df[c+str('_q3')] =(df[c].astype(np.float32 ).values > q3[c] ).astype(np.int8)
return df
df = descrictive_stat_feat(df )<feature_engineering> | def transform(image, label, DIM = image_size[0]):
XDIM = DIM % 2
rot = 15.* tf.random.normal([1],dtype='float32')
shr = 5.* tf.random.normal([1],dtype='float32')
h_zoom = 1.0 + tf.random.normal([1],dtype='float32')/10.
w_zoom = 1.0 + tf.random.normal([1],dtype='float32')/10.
h_shift = 16.* tf.random.normal([1],dtype='float32')
w_shift = 16.* tf.random.normal([1],dtype='float32')
m = get_mat(rot,shr,h_zoom,w_zoom,h_shift,w_shift)
x = tf.repeat(tf.range(DIM//2,-DIM//2,-1), DIM)
y = tf.tile(tf.range(-DIM//2,DIM//2),[DIM])
z = tf.ones([DIM*DIM],dtype='int32')
idx = tf.stack([x,y,z])
idx2 = K.dot(m,tf.cast(idx,dtype='float32'))
idx2 = K.cast(idx2,dtype='int32')
idx2 = K.clip(idx2,-DIM//2+XDIM+1,DIM//2)
idx3 = tf.stack([DIM//2-idx2[0,], DIM//2-1+idx2[1,]])
d = tf.gather_nd(image,tf.transpose(idx3))
return tf.reshape(d,[DIM,DIM,3]),label | Petals to the Metal - Flower Classification on TPU |
11,225,265 | def basic_details(df):
b = pd.DataFrame()
b['Missing value'] = df.isnull().sum()
b['N unique value'] = df.nunique()
b['dtype'] = df.dtypes
return b
basic_details(df )<prepare_x_and_y> | batch_size = 16 * strategy.num_replicas_in_sync
epochs = 25
batched_train_dataset = train_dataset.map(transform)
batched_train_dataset = batched_train_dataset.repeat()
batched_train_dataset = batched_train_dataset.shuffle(buffer_size = 2048)
batched_train_dataset = batched_train_dataset.batch(batch_size)
batched_train_dataset = batched_train_dataset.prefetch(tf.data.experimental.AUTOTUNE)
batched_val_dataset = val_dataset.batch(batch_size)
batched_val_dataset = batched_val_dataset.cache()
batched_val_dataset = batched_val_dataset.prefetch(tf.data.experimental.AUTOTUNE)
batched_test_dataset = test_dataset.batch(batch_size)
batched_test_dataset = batched_test_dataset.prefetch(tf.data.experimental.AUTOTUNE)
early_stopping = tf.keras.callbacks.EarlyStopping(monitor ='val_loss', patience= 3)
save_checkpoints = tf.keras.callbacks.ModelCheckpoint('/kaggle/working/weights.{epoch:02d}-{val_loss:.2f}.hdf5',
save_freq='epoch', period=4)
LR_START = 0.00001
LR_MAX = 0.00005 * strategy.num_replicas_in_sync
LR_MIN = 0.00001
LR_RAMPUP_EPOCHS = 5
LR_SUSTAIN_EPOCHS = 0
LR_DECAY =.8
def lr_schedule(epoch):
if epoch < LR_RAMPUP_EPOCHS:
lr =(LR_MAX - LR_START)/ LR_RAMPUP_EPOCHS * epoch + LR_START
elif epoch < LR_RAMPUP_EPOCHS + LR_SUSTAIN_EPOCHS:
lr = LR_MAX
else:
lr =(LR_MAX - LR_MIN)* LR_DECAY**(epoch - LR_RAMPUP_EPOCHS - LR_SUSTAIN_EPOCHS)+ LR_MIN
return lr
lr_callback = tf.keras.callbacks.LearningRateScheduler(lr_schedule, verbose = True ) | Petals to the Metal - Flower Classification on TPU |
11,225,265 | X_train = df[:train.shape[0]]
X_test_fin = df[train.shape[0]:]
y = train.Survived
X_train['Y'] = y
df = X_train
df.head(20)
X = df.drop('Y', axis=1)
y = df.Y<import_modules> | def get_model() :
with strategy.scope() :
feature_extractor = efficientnet.EfficientNetB7(
weights = 'noisy-student', include_top = False, input_shape=[*image_size, 3])
feature_extractor.trainable = True
model = tf.keras.Sequential([
feature_extractor,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(flower_categories), activation='softmax', dtype='float32')
])
model.compile(optimizer='adam',
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
return model | Petals to the Metal - Flower Classification on TPU |
11,225,265 | from keras.models import Sequential
from keras.layers import Dense, Activation, Dropout
import keras
from keras.optimizers import SGD
import graphviz
import eli5
from eli5.sklearn import PermutationImportance<split> | history = model.fit(
batched_train_dataset,
steps_per_epoch=num_training_samples // batch_size,
epochs=epochs,
callbacks=[lr_callback, early_stopping, save_checkpoints],
validation_data=batched_val_dataset,
verbose=2 ) | Petals to the Metal - Flower Classification on TPU |
11,225,265 | <train_model><EOS> | print("Creating submission")
test_image_dataset = batched_test_dataset.map(lambda image,idnum: image)
predictions = np.argmax(model.predict(test_image_dataset), axis = -1)
test_ids_dataset = batched_test_dataset.map(lambda image, idnum: idnum ).unbatch()
test_ids = next(iter(test_ids_dataset.batch(num_testing_samples)) ).numpy().astype('U')
np.savetxt('submission.csv', np.rec.fromarrays([test_ids, predictions]), fmt=['%s', '%d'], delimiter=',', header='id,label', comments='')
print('Done!' ) | Petals to the Metal - Flower Classification on TPU |
11,281,017 | <SOS> metric: macrofscore Kaggle data source: petals-to-the-metal-flower-classification-on-tpu<save_to_csv> | !pip install -U git+https://github.com/qubvel/efficientnet >> /dev/null | Petals to the Metal - Flower Classification on TPU |
11,281,017 | y_pred = sq.predict(X_test_fin)
y_final =(y_pred > 0.5 ).astype(int ).reshape(X_test_fin.shape[0])
sub = pd.DataFrame()
sub['PassengerId'] = test['PassengerId']
sub['Survived'] = y_final
sub['Survived'] = sub.apply(lambda r: leaks[int(r['PassengerId'])] if int(r['PassengerId'])in leaks else r['Survived'], axis=1)
sub.to_csv('submission.csv', index=False)
print("Your submission was successfully saved!")
sub.head()
<import_modules> | AUTO = tf.data.experimental.AUTOTUNE | Petals to the Metal - Flower Classification on TPU |
11,281,017 | import keras
from keras.models import Sequential
from keras.layers import Dense, Dropout, Flatten
from keras.layers import Conv2D, MaxPooling2D
from keras.utils import to_categorical
from keras.preprocessing import image
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from keras.utils import to_categorical
from tqdm import tqdm<load_from_csv> | try:
tpu = tf.distribute.cluster_resolver.TPUClusterResolver()
print('Running on TPU ', tpu.master())
except ValueError:
tpu = None
if tpu:
tf.config.experimental_connect_to_cluster(tpu)
tf.tpu.experimental.initialize_tpu_system(tpu)
strategy = tf.distribute.experimental.TPUStrategy(tpu)
else:
strategy = tf.distribute.get_strategy()
print("REPLICAS: ", strategy.num_replicas_in_sync ) | Petals to the Metal - Flower Classification on TPU |
11,281,017 | train=pd.read_csv('.. /input/train.csv')
test=pd.read_csv('.. /input/test.csv' )<prepare_x_and_y> | MIXED_PRECISION = True
XLA_ACCELERATE = True
if MIXED_PRECISION:
if tpu: policy = tf.keras.mixed_precision.experimental.Policy('mixed_bfloat16')
else: policy = tf.keras.mixed_precision.experimental.Policy('mixed_float16')
mixed_precision.set_policy(policy)
print('Mixed precision enabled')
if XLA_ACCELERATE:
tf.config.optimizer.set_jit(True)
print('Accelerated Linear Algebra enabled' ) | Petals to the Metal - Flower Classification on TPU |
11,281,017 | features = [c for c in train.columns if c not in ['label']]
y=train['label']
X = train.drop(labels = ["label"],axis = 1)
X = X.values.reshape(-1,28,28,1)
test = test.values.reshape(-1,28,28,1)
<categorify> | IMAGE_SIZES = [[512, 512], [331, 331], [224, 224], [192, 192]]
IMAGE_SIZE = IMAGE_SIZES[2]
BATCH_SIZE = 32 * strategy.num_replicas_in_sync
print('Using IMAGE_SIZE', IMAGE_SIZE, 'and BATCH_SIZE', BATCH_SIZE ) | Petals to the Metal - Flower Classification on TPU |
11,281,017 | y = to_categorical(y )<feature_engineering> | use_external = True
BASE_PATH = KaggleDatasets().get_gcs_path('tpu-getting-started')
EXT_PATH = KaggleDatasets().get_gcs_path('tf-flower-photo-tfrec')
def get_path(image_size=IMAGE_SIZE):
return f'{BASE_PATH}/tfrecords-jpeg-{image_size[0]}x{image_size[1]}'
def get_ext_path(folder, image_size=IMAGE_SIZE):
return f'{EXT_PATH}/{folder}/tfrecords-jpeg-{image_size[0]}x{image_size[1]}'
TRAINING_FILES = tf.io.gfile.glob(get_path() + '/train/*.tfrec')
VALIDATION_FILES = tf.io.gfile.glob(get_path() + '/val/*.tfrec')
TEST_FILES = tf.io.gfile.glob(get_path() + '/test/*.tfrec')
if use_external:
INATURALIST = tf.io.gfile.glob(get_ext_path('inaturalist_no_test')+ '/*.tfrec')
TRAINING_FILES += INATURALIST
OXFORD = tf.io.gfile.glob(get_ext_path('oxford_102_no_test')+ '/*.tfrec')
TRAINING_FILES += OXFORD
TFFLOWERS = tf.io.gfile.glob(get_ext_path('tf_flowers_no_test')+ '/*.tfrec')
TRAINING_FILES += TFFLOWERS
print('Loaded Training:', len(TRAINING_FILES), 'files, Validation:', len(VALIDATION_FILES), 'files, Test:', len(TEST_FILES), 'files' ) | Petals to the Metal - Flower Classification on TPU |
11,281,017 | X = X / 255.0
test = test / 255.0
<split> | CLASSES = ['pink primrose', 'hard-leaved pocket orchid', 'canterbury bells', 'sweet pea', 'wild geranium', 'tiger lily', 'moon orchid', 'bird of paradise', 'monkshood', 'globe thistle',
'snapdragon', "colt's foot", 'king protea', 'spear thistle', 'yellow iris', 'globe-flower', 'purple coneflower', 'peruvian lily', 'balloon flower', 'giant white arum lily',
'fire lily', 'pincushion flower', 'fritillary', 'red ginger', 'grape hyacinth', 'corn poppy', 'prince of wales feathers', 'stemless gentian', 'artichoke', 'sweet william',
'carnation', 'garden phlox', 'love in the mist', 'cosmos', 'alpine sea holly', 'ruby-lipped cattleya', 'cape flower', 'great masterwort', 'siam tulip', 'lenten rose',
'barberton daisy', 'daffodil', 'sword lily', 'poinsettia', 'bolero deep blue', 'wallflower', 'marigold', 'buttercup', 'daisy', 'common dandelion',
'petunia', 'wild pansy', 'primula', 'sunflower', 'lilac hibiscus', 'bishop of llandaff', 'gaura', 'geranium', 'orange dahlia', 'pink-yellow dahlia',
'cautleya spicata', 'japanese anemone', 'black-eyed susan', 'silverbush', 'californian poppy', 'osteospermum', 'spring crocus', 'iris', 'windflower', 'tree poppy',
'gazania', 'azalea', 'water lily', 'rose', 'thorn apple', 'morning glory', 'passion flower', 'lotus', 'toad lily', 'anthurium',
'frangipani', 'clematis', 'hibiscus', 'columbine', 'desert-rose', 'tree mallow', 'magnolia', 'cyclamen ', 'watercress', 'canna lily',
'hippeastrum ', 'bee balm', 'pink quill', 'foxglove', 'bougainvillea', 'camellia', 'mallow', 'mexican petunia', 'bromelia', 'blanket flower',
'trumpet creeper', 'blackberry lily', 'common tulip', 'wild rose']
print(len(CLASSES)) | Petals to the Metal - Flower Classification on TPU |
11,281,017 | X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=42, test_size=0.1 )<choose_model_class> | class DS() :
def __init__(self, files, training=True, shuffle=False, augment=False, batch_size=16, repeat=False):
self.files = files
self.augment = augment
self.ds = tf.data.TFRecordDataset(files, num_parallel_reads = AUTO)
self.ds = self.ds.map(self.read_with_label if training else self.read_without_label)
if self.augment:
self.ds = self.ds.map(self.augment_image)
if shuffle:
options = tf.data.Options()
options.experimental_deterministic = False
self.ds = self.ds.with_options(options)
self.ds = self.ds.shuffle(2048)
if repeat:
self.ds = self.ds.repeat()
self.ds = self.ds.batch(batch_size)
self.ds = self.ds.prefetch(AUTO)
def data(self):
return self.ds
def read_with_label(self, example):
example = tf.io.parse_single_example(example, {
'image': tf.io.FixedLenFeature([], tf.string),
'class': tf.io.FixedLenFeature([], tf.int64),
})
return self.decode_image(example['image']), tf.cast(example['class'], tf.int32)
def read_without_label(self, example):
example = tf.io.parse_single_example(example, {
'image': tf.io.FixedLenFeature([], tf.string),
'id': tf.io.FixedLenFeature([], tf.string),
})
return self.decode_image(example['image']), example['id']
def decode_image(self, img):
img = tf.io.decode_jpeg(img, channels=3)
img = tf.cast(img, tf.float32)/ 255.0
return tf.reshape(img, [*IMAGE_SIZE, 3])
def augment_image(self, img, label):
return Augmentor.process(img, IMAGE_SIZE[0]), label
def count(self):
n = [int(re.compile(r"-([0-9]*)\." ).search(filename ).group(1)) for filename in self.files]
return np.sum(n)
| Petals to the Metal - Flower Classification on TPU |
11,281,017 | model = Sequential()
model.add(Conv2D(64, kernel_size=(3, 3), padding='same',activation='relu',input_shape=(28,28,1)))
model.add(Conv2D(64,(3, 3), padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Conv2D(64, 5, padding='same', activation='relu'))
model.add(Conv2D(64, 5, padding='same', activation='relu'))
model.add(MaxPooling2D(( 2, 2)))
model.add(Dropout(0.5))
model.add(Flatten())
model.add(Dense(256, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(10, activation='softmax'))<choose_model_class> | items = 0
batches = 0
times = {
'batches': list() ,
'items': list() ,
'batch_speed': list() ,
'item_speed': list() ,
}
def loop(items, batches, times):
train_ds = DS(TRAINING_FILES, augment=True, batch_size=128 ).data()
start = time.time()
for element in train_ds:
so_far = time.time() - start
batches += 1
items += len(element[0])
times['batches'].append(batches)
times['items'].append(items)
times['batch_speed'].append(( batches+1)/so_far)
times['item_speed'].append(( items+1)/so_far)
print(items, so_far,(items)/so_far)
print(batches, items, so_far,(batches)/so_far,(items)/so_far)
gc.collect()
def show_graph(size):
plt.figure(figsize=(12,6))
plt.plot(times['batches'], times['batch_speed'], color='red')
plt.plot(times['items'], times['item_speed'], color='blue', linestyle='--')
plt.legend()
plt.title('Batch size' + str(size))
plt.show()
| Petals to the Metal - Flower Classification on TPU |
11,281,017 | model.compile(loss='categorical_crossentropy',optimizer='Adam',metrics=['accuracy'])
<train_model> | def test_dataset() :
print('Training images')
show_images(DS(TRAINING_FILES, augment=True, batch_size=128 ).data())
print('Validation image')
show_images(DS(VALIDATION_FILES, batch_size=128 ).data())
print('Testing images')
show_images(DS(TEST_FILES, training=False, batch_size=128 ).data())
print('Testing Augmented images')
show_images(DS(TEST_FILES, training=False, augment=True, batch_size=128 ).data())
def show_images(ds, rows=3, columns=6):
plt.figure(figsize=(12, 6))
count = 0
for i, examples in enumerate(ds.take(1)) :
images = examples[0]
labels = examples[1]
for j, image in enumerate(images):
if count == rows * columns:
break
plt.subplot(rows, columns, count+1, xticks=[], yticks=[])
plt.imshow(image)
if labels[0].dtype == tf.string:
plt.xlabel(labels[j].numpy().decode("utf-8"))
else:
plt.xlabel(CLASSES[labels[j]])
count += 1
plt.tight_layout()
plt.show()
gc.collect() | Petals to the Metal - Flower Classification on TPU |
11,281,017 | datagen = ImageDataGenerator(
rotation_range=40,
width_shift_range=0.2,
height_shift_range=0.2,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode='nearest')
datagen.fit(X_train )<train_model> | EFNS = [
efn.EfficientNetB0,
efn.EfficientNetB1,
efn.EfficientNetB2,
efn.EfficientNetB3,
efn.EfficientNetB4,
efn.EfficientNetB5,
efn.EfficientNetB6,
efn.EfficientNetB7,
]
def create_efficientnet(index, weight='imagenet'):
pretrained_model = EFNS[index](
weights = weight,
include_top = False,
input_shape = [*IMAGE_SIZE, 3]
)
pretrained_model.trainable = True
model = tf.keras.Sequential([
pretrained_model,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(1024, activation='relu', kernel_regularizer=tf.keras.regularizers.l2(0.01)) ,
tf.keras.layers.BatchNormalization() ,
tf.keras.layers.Dropout(0.5),
tf.keras.layers.Dense(len(CLASSES), activation='softmax'),
], name='efn' + str(index))
return model
def create_densenet() :
pretrained_model = tf.keras.applications.DenseNet201(
weights = 'imagenet',
include_top = False,
input_shape = [*IMAGE_SIZE, 3]
)
pretrained_model.trainable = True
model = tf.keras.Sequential([
pretrained_model,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(1024, activation='relu', kernel_regularizer=tf.keras.regularizers.l2(0.01)) ,
tf.keras.layers.BatchNormalization() ,
tf.keras.layers.Dropout(0.5),
tf.keras.layers.Dense(len(CLASSES), activation='softmax'),
], name='densenet')
return model
def create_cnn() :
model = tf.keras.Sequential([
tf.keras.layers.Conv2D(16, kernel_size=3, strides=2, padding='same', activation='relu', input_shape=[*IMAGE_SIZE, 3]),
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation='softmax')
], name='simple')
return model | Petals to the Metal - Flower Classification on TPU |
11,281,017 | batch_size = 80
results = model.fit_generator(datagen.flow(X_train, y_train, batch_size = batch_size), epochs=50,
steps_per_epoch=X_train.shape[0] // batch_size,
validation_data=(X_test, y_test))<compute_test_metric> | def compile_models(models):
for x in models:
models[x].compile(
optimizer = tf.keras.optimizers.Adam(learning_rate=1e-3),
loss = tf.keras.losses.SparseCategoricalCrossentropy() ,
metrics = ['sparse_categorical_accuracy'],
)
models[x].summary()
print('')
models = {}
with strategy.scope() :
start = time.time()
models['efn7'] = create_efficientnet(7)
print(f'Created efn7 in {(time.time() - start):.4f}s')
start = time.time()
models['efn6'] = create_efficientnet(6)
print(f'Created efn6 in {(time.time() - start):.4f}s')
start = time.time()
models['efn4-noisy'] = create_efficientnet(4, 'noisy-student')
print(f'Created efn4 in {(time.time() - start):.4f}s')
start = time.time()
models['densenet'] = create_densenet()
print(f'Created densenet in {(time.time() - start):.4f}s')
compile_models(models ) | Petals to the Metal - Flower Classification on TPU |
11,281,017 | final_loss, final_acc = model.evaluate(X_test, y_test, verbose=0)
print("Final loss: {0:.4f}, final accuracy: {1:.4f}".format(final_loss, final_acc))<save_to_csv> | train_ds = DS(TRAINING_FILES, augment=True, shuffle=True, batch_size=BATCH_SIZE, repeat=True)
val_ds = DS(VALIDATION_FILES, batch_size=BATCH_SIZE)
val_aug_ds = DS(VALIDATION_FILES, augment=True, batch_size=BATCH_SIZE)
test_ds = DS(TEST_FILES, training=False, batch_size=BATCH_SIZE)
print(f'Training: {train_ds.count() } images, Validation: {val_ds.count() }, Test: {test_ds.count() } images' ) | Petals to the Metal - Flower Classification on TPU |
11,281,017 | prediction = model.predict(test)
prediction = np.argmax(prediction,axis = 1)
output=pd.DataFrame({"ImageId": list(range(1,len(prediction)+1)) ,
"Label": prediction})
output.to_csv("output.csv", index=False, header=True )<import_modules> | lr_start = 0.00005
lr_max = 0.0000125 * strategy.num_replicas_in_sync
lr_min = 0.0000001
lr_ramp_ep = 5
lr_sus_ep = 0
lr_decay = 0.8
def lrfn(epoch):
if epoch < lr_ramp_ep:
lr =(lr_max - lr_start)/ lr_ramp_ep * epoch + lr_start
elif epoch < lr_ramp_ep + lr_sus_ep:
lr = lr_max
else:
lr =(lr_max - lr_min)* lr_decay**(epoch - lr_ramp_ep - lr_sus_ep)+ lr_min
return lr
lr_callback = tf.keras.callbacks.LearningRateScheduler(lrfn, verbose=True)
count = range(100)
plt.figure(figsize=(12, 6))
plt.subplot(1, 3, 1)
plt.plot(count, [lrfn(x)for x in count])
cosine_decay_learning_rate = tf.keras.experimental.CosineDecayRestarts(
0.001, 0.8, t_mul=2.0, m_mul=1.0, alpha=0.0,
name=None
)
| Petals to the Metal - Flower Classification on TPU |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.