kernel_id int64 24.2k 23.3M | prompt stringlengths 8 1.85M | completetion stringlengths 1 182k | comp_name stringlengths 5 57 |
|---|---|---|---|
15,318,234 | results_plabel = []
for images, image_ids in test_data_loader:
predictions = make_tta_predictions(images)
for i, image in enumerate(images):
image_id = image_ids[i]
image_ = cv2.imread(f'{DATA_ROOT_PATH}/{image_id}.jpg', cv2.IMREAD_COLOR)
h,w,_ = np.shape(image_)
boxes, scores, labels = run_wbf(predictions, image_in... | bert_layer = hub.KerasLayer('https://tfhub.dev/tensorflow/bert_en_uncased_L-12_H-768_A-12/1', trainable=True ) | Natural Language Processing with Disaster Tweets |
15,318,234 | results_df = pd.DataFrame(results_plabel, columns=['image_id', 'width','height','source','x','y','w','h'])
results_df.head()<predict_on_test> | K = 2
skf = StratifiedKFold(n_splits=K, shuffle=True ) | Natural Language Processing with Disaster Tweets |
15,318,234 | results = []
for images, image_ids in test_data_loader:
predictions = make_tta_predictions(images)
for i, image in enumerate(images):
boxes, scores, labels = run_wbf(predictions, image_index=i)
boxes =(boxes*2 ).astype(np.int32 ).clip(min=0, max=1023)
image_id = image_ids[i]
boxes[:, 2] = boxes[:, 2] - boxes[:, 0]
b... | class ClassificationReport(Callback):
def __init__(self, train_data=() , validation_data=()):
super(Callback, self ).__init__()
self.X_train, self.y_train = train_data
self.train_precision_scores = []
self.train_recall_scores = []
self.train_f1_scores = []
self.X_val, self.y_val = validation_data
self.val_precision_sco... | Natural Language Processing with Disaster Tweets |
15,318,234 | test_df = pd.DataFrame(results, columns=['image_id', 'PredictionString'])
test_df.to_csv('submission.csv', index=False )<import_modules> | class DisasterDetector:
def __init__(self, bert_layer, optimizer, max_seq_length=128, lr=0.0001, epochs=15, batch_size=32):
self.bert_layer = bert_layer
self.max_seq_length = max_seq_length
vocab_file = self.bert_layer.resolved_object.vocab_file.asset_path.numpy()
do_lower_case = self.bert_layer.resolved_object.do_lowe... | Natural Language Processing with Disaster Tweets |
15,318,234 | import numpy as np
import pandas as pd
import os
from tqdm.auto import tqdm
import shutil as sh<install_modules> | sgd = SGD(1e-3)
clf = DisasterDetector(bert_layer,sgd, max_seq_length=128, lr=0.0001, epochs=10, batch_size=32)
clf.train(train_set['text'], train_set['target'] ) | Natural Language Processing with Disaster Tweets |
15,318,234 | !cp -r.. /input/yolov5-pseudo-labeling/* .<install_modules> | y_pred = clf.predict(test_set['text'])
y_pred_thres = [1 if pred[0] >=0.5 else 0 for pred in y_pred]
y_pred_df = pd.DataFrame(y_pred_thres, columns=['target'] ) | Natural Language Processing with Disaster Tweets |
15,318,234 | !pip install --no-deps '.. /input/weightedboxesfusion/' > /dev/null<feature_engineering> | sample_subm = pd.read_csv('.. /input/nlp-getting-started/sample_submission.csv')
id = sample_subm.id | Natural Language Processing with Disaster Tweets |
15,318,234 | R_fold = 1
def convertTrainLabel() :
df = pd.read_csv('.. /input/global-wheat-detection/train.csv')
bboxs = np.stack(df['bbox'].apply(lambda x: np.fromstring(x[1:-1], sep=',')))
for i, column in enumerate(['x', 'y', 'w', 'h']):
df[column] = bboxs[:,i]
df.drop(columns=['bbox'], inplace=True)
df['x_center'] = df['x'] ... | subm = pd.concat([id, y_pred_df], axis=1 ) | Natural Language Processing with Disaster Tweets |
15,318,234 | <load_pretrained><EOS> | subm.to_csv('sample_subm.csv', index=False, header = True ) | Natural Language Processing with Disaster Tweets |
15,118,972 | <SOS> metric: meanfscore Kaggle data source: natural-language-processing-with-disaster-tweets<load_pretrained> | nltk.download('stopwords', quiet=True)
stopwords = stopwords.words('english')
sns.set(style="white", font_scale=1.2)
plt.rcParams["figure.figsize"] = [10,8]
pd.set_option.display_max_columns = 0
pd.set_option.display_max_rows = 0 | Natural Language Processing with Disaster Tweets |
15,118,972 |
<define_variables> | train = pd.read_csv(".. /input/nlp-getting-started/train.csv")
test = pd.read_csv(".. /input/nlp-getting-started/test.csv" ) | Natural Language Processing with Disaster Tweets |
15,118,972 | def format_prediction_string(boxes, scores):
pred_strings = []
for j in zip(scores, boxes):
pred_strings.append("{0:.4f} {1} {2} {3} {4}".format(j[0], j[1][0], j[1][1], j[1][2], j[1][3]))
return " ".join(pred_strings )<import_modules> | null_counts = pd.DataFrame({"Num_Null": train.isnull().sum() })
null_counts["Pct_Null"] = null_counts["Num_Null"] / train.count() * 100
null_counts | Natural Language Processing with Disaster Tweets |
15,118,972 | from utils.datasets import *
from utils.utils import *<train_on_grid> | len(train["keyword"].value_counts() ) | Natural Language Processing with Disaster Tweets |
15,118,972 | def detect1Image_aug(im0, imgsz, model, device, conf_thres, iou_thres):
img = letterbox(im0, new_shape=imgsz)[0]
img = img[:, :, ::-1].transpose(2, 0, 1)
img = np.ascontiguousarray(img)
img = torch.from_numpy(img ).to(device)
img = img.float()
img /= 255.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
pred = mod... | def keyword_disaster_probabilities(x):
tweets_w_keyword = np.sum(train["keyword"].fillna("" ).str.contains(x))
tweets_w_keyword_disaster = np.sum(train["keyword"].fillna("" ).str.contains(x)& train["target"] == 1)
return tweets_w_keyword_disaster / tweets_w_keyword
keywords_vc["Disaster_Probability"] = keywords_vc.ind... | Natural Language Processing with Disaster Tweets |
15,118,972 | def clip_coords2(boxes, img_shape):
boxes[:, 0].clamp_(0, img_shape[1])
boxes[:, 1].clamp_(0, img_shape[0])
boxes[:, 2].clamp_(0, img_shape[1])
boxes[:, 3].clamp_(0, img_shape[0])
def scale_coords2(coords,factorx,factory, img0_shape):
coords[:, 0::2] *= factorx
coords[:, 1::2] *= factory
clip_coords2(coords, img0_s... | keywords_vc.sort_values(by="Disaster_Probability", ascending=False ).head(10 ) | Natural Language Processing with Disaster Tweets |
15,118,972 | !pip install --no-deps '.. /input/timm-package/timm-0.1.26-py3-none-any.whl' > /dev/null
!pip install --no-deps '.. /input/pycocotools/pycocotools-2.0-cp37-cp37m-linux_x86_64.whl' > /dev/null<install_modules> | len(train["location"].value_counts() ) | Natural Language Processing with Disaster Tweets |
15,118,972 | !pip install /kaggle/input/orkatzfdata/yacs-0.1.7-py3-none-any.whl
!mkdir fvcore
!cp -R '/kaggle/input/orkatzfdata/fvcore-0.1.dev200407/fvcore-0.1.dev200407/'./fvcore
!pip install fvcore/fvcore-0.1.dev200407/.
!mkdir detectron2-ResNeSt
!cp -R /kaggle/input/orkatzfdata/detectron2-ResNeSt/*./detectron2-ResNeSt/
!pip ins... | def create_corpus(target):
corpus = []
for w in train.loc[train["target"] == target]["text"].str.split() :
for i in w:
corpus.append(i)
return corpus
def create_corpus_dict(target):
corpus = create_corpus(target)
stop_dict = defaultdict(int)
for word in corpus:
if word in stopwords:
stop_dict[word] += 1
return sorte... | Natural Language Processing with Disaster Tweets |
15,118,972 | def load_net(checkpoint_path):
config = get_efficientdet_config('tf_efficientdet_d5')
net = EfficientDet(config, pretrained_backbone=False)
config.num_classes = 1
config.image_size = 512
net.class_net = HeadNet(config, num_outputs=config.num_classes, norm_kwargs=dict(eps=.001, momentum=.01))
checkpoint = torch.load(c... | corpus_disaster, corpus_non_disaster = create_corpus(1), create_corpus(0)
counter_disaster, counter_non_disaster = Counter(corpus_disaster), Counter(corpus_non_disaster)
x_disaster, y_disaster, x_non_disaster, y_non_disaster = [], [], [], []
counter = 0
for word, count in counter_disaster.most_common() [0:100]:
if(wo... | Natural Language Processing with Disaster Tweets |
15,118,972 | models2 = [load_net7('.. /input/tempb7/best-checkpoint-015epoch.bin'),
load_net7('.. /input/tempb7/best-checkpoint-020epoch.bin'),
load_net7('.. /input/tempb7/best-checkpoint-022epoch.bin'),]
models = [
load_net('.. /input/effdetbestpth/best-fold0-augmix.pth'),
load_net('.. /input/effdetbestpth/best-fold3.pth'),
load_n... | def bigrams(target):
corpus = train[train["target"] == target]["text"]
count_vec = CountVectorizer(ngram_range=(2, 2)).fit(corpus)
bag_of_words = count_vec.transform(corpus)
sum_words = bag_of_words.sum(axis=0)
words_freq = [(word, sum_words[0, idx])for word, idx in count_vec.vocabulary_.items() ]
words_freq =sorted... | Natural Language Processing with Disaster Tweets |
15,118,972 | cfg = get_cfg()
cfg.merge_from_file(model_zoo.get_config_file("COCO-Detection/faster_cascade_rcnn_ResNeSt_101_FPN_syncbn_range-scale_1x.yaml"))
cfg.MODEL.ROI_HEADS.NUM_CLASSES = 1
cfg.MODEL.WEIGHTS = os.path.join('/kaggle/input/best-inrae-1/', "model_final.pth")
cfg.MODEL.ROI_HEADS.SCORE_THRESH_TEST = 0.47
cfg.DATASET... | def remove_pattern(input_txt, pattern):
r = re.findall(pattern, input_txt)
for i in r:
input_txt = re.sub(i, '', input_txt)
return input_txt
train['tweet'] = np.vectorize(remove_pattern )(train['text'], "
test['tweet'] = np.vectorize(remove_pattern )(test['text'], "
train.head()
train['tweet'] = train['tweet'].str.re... | Natural Language Processing with Disaster Tweets |
15,118,972 | models3 =[predictor1]<data_type_conversions> | warnings.filterwarnings("ignore")
tqdm.pandas()
stopword=set(STOPWORDS)
lem = WordNetLemmatizer()
tokenizer=TweetTokenizer()
np.random.seed(0)
random_state = 29 | Natural Language Processing with Disaster Tweets |
15,118,972 | DATA_ROOT_PATH = '.. /input/global-wheat-detection/test'
class TestDatasetRetriever(Dataset):
def __init__(self, image_ids, transforms=None,transforms2=None):
super().__init__()
self.image_ids = image_ids
self.transforms = transforms
self.transforms2 = transforms2
def __getitem__(self, index: int):
image_id = self.imag... | !pip install GPUtil
def free_gpu_cache() :
print("Initial GPU Usage")
gpu_usage()
torch.cuda.empty_cache()
cuda.select_device(0)
cuda.close()
cuda.select_device(0)
for obj in gc.get_objects() :
if torch.is_tensor(obj):
del obj
gc.collect()
print("GPU Usage after emptying the cache")
gpu_usage() | Natural Language Processing with Disaster Tweets |
15,118,972 | def make_predictions(
images, images1,image_ids,
score_threshold=0.25,
):
images = images.cuda().float()
images1 = images1.cuda().float()
image_id = image_ids
rh,rw,_ = cv2.imread(f'{DATA_ROOT_PATH}/{image_id}.jpg' ).shape
Hscale512 = rh/512
Wscale512 = rw/512
Hscale1024 = rh/1024
Wscale1024 = rw/1024
predictions = [... | train = pd.read_csv(".. /input/nlp-getting-started/train.csv")
test = pd.read_csv(".. /input/nlp-getting-started/test.csv")
sub= pd.read_csv(".. /input/nlp-getting-started/sample_submission.csv" ) | Natural Language Processing with Disaster Tweets |
15,118,972 | def clip_coords3(boxes, img_shape):
boxes[:, 0].clamp_(0, img_shape[1])
boxes[:, 1].clamp_(0, img_shape[0])
boxes[:, 2].clamp_(0, img_shape[1])
boxes[:, 3].clamp_(0, img_shape[0])
return boxes<compute_test_metric> | abbreviations = {
"$" : " dollar ",
"€" : " euro ",
"4ao" : "for adults only",
"a.m" : "before midday",
"a3" : "anytime anywhere anyplace",
"aamof" : "as a matter of fact",
"acct" : "account",
"adih" : "another day in hell",
"afaic" : "as far as i am concerned",
"afaict" : "as far as i can tell",
"afaik" : "as far as i... | Natural Language Processing with Disaster Tweets |
15,118,972 | def run_wbf2(predictions, image_index, image_size=1024, iou_thr=0.34, skip_box_thr=0.33, weights=None):
boxes = [(prediction[image_index]['boxes']/(image_size-1)).tolist() for prediction in predictions]
scores = [prediction[image_index]['scores'].tolist() for prediction in predictions]
labels = [np.ones(prediction[imag... | def remove_URL(text):
url = re.compile(r'https?://\S+|www\.\S+')
return url.sub(r'URL',text)
def remove_HTML(text):
html=re.compile(r'<.*?>')
return html.sub(r'',text)
def remove_not_ASCII(text):
text = ''.join([word for word in text if word in string.printable])
return text
def word_abbrev(word):
return abbreviat... | Natural Language Processing with Disaster Tweets |
15,118,972 | def detect() :
transforms = get_valid_transforms()
transforms2 = get_valid_transforms2()
source = '.. /input/global-wheat-detection/test/'
weights = 'weights/best.pt'
weights800 = '.. /input/yolo800/best_yolov5x_fold0_800.pt'
if not os.path.exists(weights):
weights = '.. /input/yolov5pth/weightsbest_yolov5x_fold3.pt'
i... | def clean_tweet(text):
text = remove_URL(text)
text = remove_HTML(text)
text = remove_not_ASCII(text)
text = text.lower()
text = replace_abbrev(text)
text = remove_mention(text)
text = remove_number(text)
text = remove_emoji(text)
text = transcription_sad(text)
text = transcription_smile(text)
text = transcrip... | Natural Language Processing with Disaster Tweets |
15,118,972 | results = detect()
test_df = pd.DataFrame(results, columns=['image_id', 'PredictionString'])
test_df.to_csv('submission.csv', index=False)
test_df.head()<install_modules> | train["clean_text"] = train["text"].apply(clean_tweet)
test["clean_text"] = test["text"].apply(clean_tweet)
train["clean_tokens"] = train["clean_text"].apply(lambda x: word_tokenize(x))
test["clean_tokens"] = test["clean_text"].apply(lambda x: word_tokenize(x)) | Natural Language Processing with Disaster Tweets |
15,118,972 | ! pip install --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext".. /input/nvidiaapex/<install_modules> | skip_gram_model = Word2Vec(train['clean_tokens'],size=150,window=3,min_count=2,sg=1)
skip_gram_model.train(train['clean_tokens'],total_examples=len(train['clean_tokens']),epochs=10)
cbow_model = Word2Vec(train['clean_tokens'],size=150,window=3,min_count=2)
cbow_model.train(train['clean_tokens'],total_examples=len(tr... | Natural Language Processing with Disaster Tweets |
15,118,972 | !pip install --no-deps '.. /input/timm-package/timm-0.1.26-py3-none-any.whl' > /dev/null
!pip install --no-deps '.. /input/pycocotools/pycocotools-2.0-cp37-cp37m-linux_x86_64.whl' > /dev/null<define_variables> | max_features=5000
count_vectorizer = CountVectorizer(max_features=max_features)
sparce_matrix_train=count_vectorizer.fit_transform(train['clean_text'])
sparce_matrix_test=count_vectorizer.fit_transform(train['clean_text'])
def count_vector(data):
count_vectorizer = CountVectorizer()
vect = count_vectorizer.fit_trans... | Natural Language Processing with Disaster Tweets |
15,118,972 | look_at_on_kernel = 1<set_options> | metrics = pd.DataFrame(columns=['model' ,'vectoriser', 'f1 score', 'train accuracy','test accuracy'] ) | Natural Language Processing with Disaster Tweets |
15,118,972 | SEED = 42
def seed_everything(seed):
random.seed(seed)
os.environ['PYTHONHASHSEED'] = str(seed)
np.random.seed(seed)
torch.manual_seed(seed)
torch.cuda.manual_seed(seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = True
seed_everything(SEED )<load_from_csv> | models=[
XGBClassifier(max_depth=6, n_estimators=1000),
LogisticRegression(random_state=random_state),
SVC(random_state=random_state),
MultinomialNB() ,
DecisionTreeClassifier(random_state = random_state),
KNeighborsClassifier() ,
RandomForestClassifier(random_state=random_state),
] | Natural Language Processing with Disaster Tweets |
15,118,972 | marking = pd.read_csv('.. /input/pure-box/cleanedTrainNoIndexOnLimit.csv' )<define_search_model> | for model in models:
y = train.target
x = X_train_count
x_train, x_test, y_train, y_test = train_test_split(x,y, test_size = 0.3)
fit_and_predict(model,x_train,x_test,y_train,y_test,'Count vector')
x = X_train_tfidf
x_train, x_test, y_train, y_test = train_test_split(x,y, test_size = 0.3)
fit_and_predict(model,x_tra... | Natural Language Processing with Disaster Tweets |
15,118,972 | def get_train_transforms() :
return A.Compose(
[
A.RandomSizedCrop(min_max_height=(800, 800), height=1024, width=1024, p=0.5),
A.OneOf([
A.HueSaturationValue(hue_shift_limit=0.2, sat_shift_limit= 0.2,
val_shift_limit=0.2, p=0.9),
A.RandomBrightnessContrast(brightness_limit=0.2,
contrast_limit=0.2, p=0.9),
],p=0.9),
A.... | metrics = metrics.sort_values('f1 score',ascending=False ) | Natural Language Processing with Disaster Tweets |
15,118,972 | warnings.filterwarnings("ignore")
class Fitter:
def __init__(self, model, device, config):
self.config = config
self.epoch = 0
self.base_dir = f'./{config.folder}'
if not os.path.exists(self.base_dir):
os.makedirs(self.base_dir)
self.log_path = f'{self.base_dir}/log.txt'
self.best_summary_loss = 10**5
self.model = mo... | free_gpu_cache() | Natural Language Processing with Disaster Tweets |
15,118,972 | class TrainGlobalConfig:
num_workers = 2
batch_size = 1
if apex_on:
batch_size *= 2
n_epochs = 3
lr = 0.0001
folder = 'plabel_model'
verbose = True
verbose_step = 1
step_scheduler = False
validation_scheduler = True
SchedulerClass = torch.optim.lr_scheduler.ReduceLROnPlateau
scheduler_params = dict(
mode='min',
factor... | from tensorflow.keras.preprocessing.text import Tokenizer
from tensorflow.keras.preprocessing.sequence import pad_sequences
from tensorflow import keras
from keras.models import Sequential
from keras.layers import Dense, Embedding, LSTM,GRU, Dropout, Activation, Input, Flatten, Bidirectional, Conv1D, MaxPooling1D
from ... | Natural Language Processing with Disaster Tweets |
15,118,972 | DATA_ROOT_PATH = '.. /input/global-wheat-detection/test'
class DatasetRetriever(Dataset):
def __init__(self, image_ids, transforms=None):
super().__init__()
self.image_ids = image_ids
self.transforms = transforms
def __getitem__(self, index: int):
image_id = self.image_ids[index]
image = cv2.imread(f'{DATA_ROOT_PATH}/{... | def train_lstm(x_train,x_test,y_train,y_test,vectorizer_name,vocab_size,input_length):
epochs = 1
verbose = 1
batch_size = 32
embed_dim = 32
optimizer = optimizers.Adam(lr=0.002)
model = Sequential()
model.add(Embedding(vocab_size, embed_dim,input_length = input_length))
model.add(Dropout(0.2))
model.add(LSTM(32, drop... | Natural Language Processing with Disaster Tweets |
15,118,972 | dataset = DatasetRetriever(
image_ids=np.array([path.split('/')[-1][:-4] for path in glob(f'{DATA_ROOT_PATH}/*.jpg')]),
transforms=get_test_transforms()
)
def collate_fn(batch):
return tuple(zip(*batch))
data_loader = DataLoader(
dataset,
batch_size=1,
shuffle=False,
num_workers=0,
drop_last=False,
collate_fn=colla... | y = train['target'].values
x_train, x_test, y_train, y_test = train_test_split(X_train_skip_gram,y, test_size = 0.3)
train_lstm(x_train,x_test,y_train,y_test, 'skip gram vector',5329,150)
| Natural Language Processing with Disaster Tweets |
15,118,972 | def load_net(checkpoint_path):
config = get_efficientdet_config('tf_efficientdet_d5')
net = EfficientDet(config, pretrained_backbone=False)
config.num_classes = 1
config.image_size=1024
net.class_net = HeadNet(config, num_outputs=config.num_classes, norm_kwargs=dict(eps=.001, momentum=.01))
checkpoint = torch.load(ch... | %reset -f | Natural Language Processing with Disaster Tweets |
15,118,972 | class BaseWheatTTA:
image_size = 1024
def augment(self, image):
raise NotImplementedError
def batch_augment(self, images):
raise NotImplementedError
def deaugment_boxes(self, boxes):
raise NotImplementedError
class TTAHorizontalFlip(BaseWheatTTA):
def augment(self, image):
return image.flip(1)
def batch_augment(se... | !pip install GPUtil
def free_gpu_cache() :
print("Initial GPU Usage")
gpu_usage()
torch.cuda.empty_cache()
cuda.select_device(0)
cuda.close()
cuda.select_device(0)
for obj in gc.get_objects() :
if torch.is_tensor(obj):
del obj
gc.collect()
print("GPU Usage after emptying the cache")
gpu_usage()
free_gpu_cache() | Natural Language Processing with Disaster Tweets |
15,118,972 | tta_transforms = []
for tta_combination in product([TTAHorizontalFlip() , None],
[TTAVerticalFlip() , None],
[TTARotate90() , None]):
tta_transforms.append(TTACompose([tta_transform for tta_transform in tta_combination if tta_transform]))<categorify> | import re
import torch
from transformers import ElectraTokenizer, ElectraForSequenceClassification,AdamW
import torch
from sklearn.metrics import classification_report
import random
import time
import datetime
import numpy as np
import pandas as pd
from transformers import get_linear_schedule_with_warmup
from torch.uti... | Natural Language Processing with Disaster Tweets |
15,118,972 | def make_tta_predictions(images, score_threshold=0.01):
with torch.no_grad() :
images = torch.stack(images ).float().cuda()
predictions = []
for tta_transform in tta_transforms:
result = []
det = net(tta_transform.batch_augment(images.clone()), torch.tensor([1]*images.shape[0] ).float().cuda())
for i in range(images.s... | if torch.cuda.is_available() :
device = torch.device("cuda")
print('We will use the GPU:', torch.cuda.get_device_name(0))
else:
print('No GPU available, using the CPU instead.')
device = torch.device("cpu" ) | Natural Language Processing with Disaster Tweets |
15,118,972 | def format_prediction_string(boxes, scores):
pred_strings = []
for j in zip(scores, boxes):
pred_strings.append("{0:.4f} {1} {2} {3} {4}".format(j[0], j[1][0], j[1][1], j[1][2], j[1][3]))
return " ".join(pred_strings )<categorify> | train = pd.read_csv(".. /input/nlp-getting-started/train.csv")
test = pd.read_csv(".. /input/nlp-getting-started/test.csv")
df_train= train
df_test= test | Natural Language Processing with Disaster Tweets |
15,118,972 | results_plabel = []
for images, image_ids in data_loader:
predictions = make_tta_predictions(images)
for i, image in enumerate(images):
image_id = image_ids[i]
image_ = cv2.imread(f'{DATA_ROOT_PATH}/{image_id}.jpg', cv2.IMREAD_COLOR)
h,w,_ = np.shape(image_)
boxes, scores, labels = run_wbf(predictions, image_index=i... | def preprocess(text):
text=text.lower()
text = re.sub(r'https?:\/\/.*[\r
]*', '', text)
text = re.sub(r'http?:\/\/.*[\r
]*', '', text)
text=text.replace(r'&?',r'and')
text=text.replace(r'<',r'<')
text=text.replace(r'>',r'>')
text = re.sub(r"(?:\@)\w+", '', text)
text=text.encode("ascii",errors="ignore" ... | Natural Language Processing with Disaster Tweets |
15,118,972 | results_df = pd.DataFrame(results_plabel, columns=['image_id', 'width','height','source','x','y','w','h'])
results_df.head()<feature_engineering> | df_train=df_train[["text","target"]] | Natural Language Processing with Disaster Tweets |
15,118,972 | results_df['image_id'] = results_df['image_id'].apply(lambda x: DATA_ROOT_PATH+'/'+ x+'.jpg' )<feature_engineering> | texts = df_train.text.values
labels = df_train.target.values | Natural Language Processing with Disaster Tweets |
15,118,972 | TRAIN_ROOT_PATH = '.. /input/global-wheat-detection/train'
marking['image_id'] = marking['image_id'].apply(lambda x: TRAIN_ROOT_PATH+'/'+ x+'.jpg' )<concatenate> | torch.cuda.empty_cache()
tokenizer = ElectraTokenizer.from_pretrained('google/electra-base-discriminator')
model = ElectraForSequenceClassification.from_pretrained('google/electra-base-discriminator',num_labels=2)
model.cuda() | Natural Language Processing with Disaster Tweets |
15,118,972 | if len(os.listdir('.. /input/global-wheat-detection/test/')) <11:
train_data_plabel = results_df
else:
train_data_plabel = pd.concat([results_df, marking], axis=0 )<feature_engineering> | indices=tokenizer.batch_encode_plus(texts,max_length=64,add_special_tokens=True, return_attention_mask=True,pad_to_max_length=True,truncation=True)
input_ids=indices["input_ids"]
attention_masks=indices["attention_mask"] | Natural Language Processing with Disaster Tweets |
15,118,972 | skf = StratifiedKFold(n_splits=5, shuffle=True, random_state=42)
df_folds = train_data_plabel[['image_id']].copy()
df_folds.loc[:, 'bbox_count'] = 1
df_folds = df_folds.groupby('image_id' ).count()
df_folds.loc[:, 'source'] = marking[['image_id', 'source']].groupby('image_id' ).min() ['source']
df_folds.loc[:, 'strati... | train_inputs, validation_inputs, train_labels, validation_labels = train_test_split(input_ids, labels,
random_state=42, test_size=0.2)
train_masks, validation_masks, _, _ = train_test_split(attention_masks, labels,
random_state=42, test_size=0.2 ) | Natural Language Processing with Disaster Tweets |
15,118,972 | TRAIN_ROOT_PATH = '.. /input/global-wheat-detection/train'
class DatasetRetriever(Dataset):
def __init__(self, marking, image_ids, transforms=None, test=False):
super().__init__()
self.image_ids = image_ids
self.marking = marking
self.transforms = transforms
self.test = test
def __getitem__(self, index: int):
image_id ... | train_inputs = torch.tensor(train_inputs)
validation_inputs = torch.tensor(validation_inputs)
train_labels = torch.tensor(train_labels, dtype=torch.long)
validation_labels = torch.tensor(validation_labels, dtype=torch.long)
train_masks = torch.tensor(train_masks, dtype=torch.long)
validation_masks = torch.tensor(v... | Natural Language Processing with Disaster Tweets |
15,118,972 | fold_number = 0
train_dataset = DatasetRetriever(
image_ids=df_folds[df_folds['fold'] != fold_number].index.values,
marking=train_data_plabel,
transforms=get_train_transforms() ,
test=False,
)
validation_dataset = DatasetRetriever(
image_ids=df_folds[df_folds['fold'] == fold_number].index.values,
marking=train_data... | batch_size = 32
train_data = TensorDataset(train_inputs, train_masks, train_labels)
train_sampler = RandomSampler(train_data)
train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=batch_size)
validation_data = TensorDataset(validation_inputs, validation_masks, validation_labels)
validation_sam... | Natural Language Processing with Disaster Tweets |
15,118,972 | def collate_fn(batch):
return tuple(zip(*batch))
def run_training() :
device = torch.device('cuda:0')
net.to(device)
train_loader = torch.utils.data.DataLoader(
train_dataset,
batch_size=TrainGlobalConfig.batch_size,
sampler=RandomSampler(train_dataset),
pin_memory=False,
drop_last=True,
num_workers=TrainGlobalConfi... | optimizer = AdamW(model.parameters() ,
lr = 6e-6,
eps = 1e-8
)
epochs = 5
total_steps = len(train_dataloader)* epochs
scheduler = get_linear_schedule_with_warmup(optimizer,
num_warmup_steps = 0,
num_training_steps = total_steps ) | Natural Language Processing with Disaster Tweets |
15,118,972 | def get_net() :
config = get_efficientdet_config('tf_efficientdet_d5')
net = EfficientDet(config, pretrained_backbone=False)
config.num_classes = 1
config.image_size = 1024
net.class_net = HeadNet(config, num_outputs=config.num_classes, norm_kwargs=dict(eps=.001, momentum=.01))
checkpoint = torch.load('.. /input/weig... | def flat_accuracy(preds, labels):
pred_flat = np.argmax(preds, axis=1 ).flatten()
labels_flat = labels.flatten()
return np.sum(pred_flat == labels_flat)/ len(labels_flat ) | Natural Language Processing with Disaster Tweets |
15,118,972 | class AverageMeter(object):
def __init__(self):
self.reset()
def reset(self):
self.val = 0
self.avg = 0
self.sum = 0
self.count = 0
def update(self, val, n=1):
self.val = val
self.sum += val * n
self.count += n
self.avg = self.sum / self.count<train_model> | seed_val = 42
random.seed(seed_val)
np.random.seed(seed_val)
torch.manual_seed(seed_val)
torch.cuda.manual_seed_all(seed_val)
loss_values = []
for epoch_i in range(0, epochs):
print("")
print('======== Epoch {:} / {:} ========'.format(epoch_i + 1, epochs))
print('Training...')
t0 = time.time()
total_loss = 0
mode... | Natural Language Processing with Disaster Tweets |
15,118,972 | if len(os.listdir('.. /input/global-wheat-detection/test/')) <1:
pass
else:
run_training()<set_options> | print("")
print("Running Validation...")
t0 = time.time()
model.eval()
preds=[]
true=[]
eval_loss, eval_accuracy = 0, 0
nb_eval_steps, nb_eval_examples = 0, 0
for batch in validation_dataloader:
batch = tuple(t.to(device)for t in batch)
b_input_ids, b_input_mask, b_labels = batch
with torch.no_grad() :
outputs = mod... | Natural Language Processing with Disaster Tweets |
15,118,972 | time.sleep(1)
def memory_cleanup() :
for obj in gc.get_objects() :
if torch.is_tensor(obj):
del obj
gc.collect()
torch.cuda.empty_cache()
memory_cleanup()<categorify> | report = {}
report['model'] = 'Electra'
report['test accuracy'] = 0.82
metrics = metrics.append(report,ignore_index=True ) | Natural Language Processing with Disaster Tweets |
15,118,972 | results_plabel = []
for images, image_ids in data_loader:
predictions = make_tta_predictions(images)
for i, image in enumerate(images):
image_id = image_ids[i]
image_ = cv2.imread(f'{DATA_ROOT_PATH}/{image_id}.jpg', cv2.IMREAD_COLOR)
h,w,_ = np.shape(image_)
boxes, scores, labels = run_wbf(predictions, image_index=i... | flat_predictions = [item for sublist in preds for item in sublist]
flat_predictions = np.argmax(flat_predictions, axis=1 ).flatten()
flat_true_labels = [item for sublist in true for item in sublist] | Natural Language Processing with Disaster Tweets |
15,118,972 | results_df = pd.DataFrame(results_plabel, columns=['image_id', 'width','height','source','x','y','w','h'])
results_df.head()<feature_engineering> | comments1 = df_test.text.values
indices1=tokenizer.batch_encode_plus(comments1,max_length=128,add_special_tokens=True, return_attention_mask=True,pad_to_max_length=True,truncation=True)
input_ids1=indices1["input_ids"]
attention_masks1=indices1["attention_mask"]
prediction_inputs1= torch.tensor(input_ids1)
prediction... | Natural Language Processing with Disaster Tweets |
15,118,972 | results_df['image_id'] = results_df['image_id'].apply(lambda x: DATA_ROOT_PATH+'/'+ x+'.jpg' )<define_variables> | print('Predicting labels for {:,} test sentences...'.format(len(prediction_inputs1)))
model.eval()
predictions = []
for batch in prediction_dataloader1:
batch = tuple(t.to(device)for t in batch)
b_input_ids1, b_input_mask1 = batch
with torch.no_grad() :
outputs1 = model(b_input_ids1, token_type_ids=None,
attention_ma... | Natural Language Processing with Disaster Tweets |
15,118,972 | TRAIN_ROOT_PATH = '.. /input/global-wheat-detection/train'
<concatenate> | sample_sub=pd.read_csv('.. /input/nlp-getting-started/sample_submission.csv')
submit=pd.DataFrame({'id':sample_sub['id'].values.tolist() ,'target':flat_predictions} ) | Natural Language Processing with Disaster Tweets |
15,118,972 | <feature_engineering><EOS> | df_leak = pd.read_csv('/kaggle/input/disasters-on-social-media/socialmedia-disaster-tweets-DFE.csv', encoding ='ISO-8859-1')[['choose_one', 'text']]
df_leak['target'] =(df_leak['choose_one'] == 'Relevant' ).astype(np.int8)
df_leak['id'] = df_leak.index.astype(np.int16)
df_leak.drop(columns=['choose_one', 'text'], inp... | Natural Language Processing with Disaster Tweets |
14,922,728 | <SOS> metric: meanfscore Kaggle data source: natural-language-processing-with-disaster-tweets<create_dataframe> | import numpy as np
import torch
from torch.utils.data import DataLoader, Dataset, TensorDataset
import sys
import torch.nn as nn
import torch.nn.functional as F
from torch.utils import data
import torch.optim as optim
import seaborn as sns
from collections import defaultdict
import time
import pandas as pd
import matpl... | Natural Language Processing with Disaster Tweets |
14,922,728 | fold_number = 0
train_dataset = DatasetRetriever(
image_ids=df_folds[df_folds['fold'] != fold_number].index.values,
marking=train_data_plabel,
transforms=get_train_transforms() ,
test=False,
)
validation_dataset = DatasetRetriever(
image_ids=df_folds[df_folds['fold'] == fold_number].index.values,
marking=train_data... | device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')
print(device ) | Natural Language Processing with Disaster Tweets |
14,922,728 | def collate_fn(batch):
return tuple(zip(*batch))
def run_training() :
device = torch.device('cuda:0')
net.to(device)
train_loader = torch.utils.data.DataLoader(
train_dataset,
batch_size=TrainGlobalConfig.batch_size,
sampler=RandomSampler(train_dataset),
pin_memory=False,
drop_last=True,
num_workers=TrainGlobalConfi... |
train_csv = pd.read_csv('/kaggle/input/nlp-getting-started/train.csv', keep_default_na = False)
train_csv = train_csv.sample(frac=1 ).reset_index(drop=True)
ninetyfive_percent = round(0.90*(len(train_csv)))
train_data = train_csv.iloc[:ninetyfive_percent]
valid_data = train_csv.iloc[ninetyfive_percent:]
print('Num... | Natural Language Processing with Disaster Tweets |
14,922,728 | def get_net() :
config = get_efficientdet_config('tf_efficientdet_d5')
net = EfficientDet(config, pretrained_backbone=False)
config.num_classes = 1
config.image_size=1024
net.class_net = HeadNet(config, num_outputs=config.num_classes, norm_kwargs=dict(eps=.001, momentum=.01))
checkpoint = torch.load('plabel_model/las... |
test_csv = pd.read_csv('/kaggle/input/nlp-getting-started/test.csv', keep_default_na = False)
test_dataset = mydataset(test_csv , name = 'test')
test_dataloader = data.DataLoader(test_dataset, shuffle= False, batch_size = 1, num_workers=16,pin_memory=True ) | Natural Language Processing with Disaster Tweets |
14,922,728 | if len(os.listdir('.. /input/global-wheat-detection/test/')) <1:
pass
else:
run_training()<set_options> | def train(model, data_loader, valid_loader, criterion, optimizer, lr_scheduler, modelpath, device, epochs):
model.train()
train_loss= []
valid_loss = []
valid_acc = []
for epoch in range(epochs):
avg_loss = 0.0
for batch_num,(tweet, input_id, attention_masks, target)in enumerate(data_loader):
input_ids, attention_masks... | Natural Language Processing with Disaster Tweets |
14,922,728 | time.sleep(1)
def memory_cleanup() :
for obj in gc.get_objects() :
if torch.is_tensor(obj):
del obj
gc.collect()
torch.cuda.empty_cache()
memory_cleanup()<categorify> | modelname = 'BERT'
modelpath = 'saved_checkpoint_'+modelname
train_loss, valid_loss, valid_acc = train(model, train_dataloader, validation_dataloader, criterion, optimizer, lr_scheduler, modelpath, device, epochs = num_Epochs ) | Natural Language Processing with Disaster Tweets |
14,922,728 | def get_valid_transforms() :
return A.Compose([
A.Resize(height=1024, width=1024, p=1.0),
ToTensorV2(p=1.0),
], p=1.0 )<data_type_conversions> | def predict(model, test_loader, device):
model.eval()
target = []
for batch_num,(captions, input_id, attention_masks)in enumerate(test_loader):
input_ids, attention_masks = input_id.to(device), attention_masks.to(device)
output_dictionary = model(input_ids,
token_type_ids=None,
attention_mask=attention_masks,
return_d... | Natural Language Processing with Disaster Tweets |
14,922,728 | <load_pretrained><EOS> | predict(model, test_dataloader, device ) | Natural Language Processing with Disaster Tweets |
14,958,171 | <SOS> metric: meanfscore Kaggle data source: natural-language-processing-with-disaster-tweets<categorify> | !pip install pytorch-pretrained-bert pytorch-nlp | Natural Language Processing with Disaster Tweets |
14,958,171 | class BaseWheatTTA:
image_size = 1024
def augment(self, image):
raise NotImplementedError
def batch_augment(self, images):
raise NotImplementedError
def deaugment_boxes(self, boxes):
raise NotImplementedError
class TTAHorizontalFlip(BaseWheatTTA):
def augment(self, image):
return image.flip(1)
def batch_augment(se... | warnings.filterwarnings('ignore')
| Natural Language Processing with Disaster Tweets |
14,958,171 | def process_det(index, det, score_threshold=0.25):
boxes = det[index].detach().cpu().numpy() [:,:4]
scores = det[index].detach().cpu().numpy() [:,4]
boxes[:, 2] = boxes[:, 2] + boxes[:, 0]
boxes[:, 3] = boxes[:, 3] + boxes[:, 1]
boxes =(boxes ).clip(min=0, max=1023 ).astype(int)
indexes = np.where(scores>score_thresho... | nltk.download('punkt' ) | Natural Language Processing with Disaster Tweets |
14,958,171 | tta_transforms = []
for tta_combination in product([TTAHorizontalFlip() , None],
[TTAVerticalFlip() , None],
[TTARotate90() , TTARotate180() , TTARotate270() , None]):
tta_transforms.append(TTACompose([tta_transform for tta_transform in tta_combination if tta_transform]))<categorify> | pd.options.display.max_colwidth = 100
seed_val=42
tf.random.set_seed(seed_val)
random.seed(seed_val)
np.random.seed(seed_val)
torch.manual_seed(seed_val)
torch.cuda.manual_seed_all(seed_val ) | Natural Language Processing with Disaster Tweets |
14,958,171 | def make_tta_predictions(images, score_threshold=0.25):
with torch.no_grad() :
images = torch.stack(images ).float().cuda()
predictions = []
for tta_transform in tta_transforms:
result = []
det = net(tta_transform.batch_augment(images.clone()), torch.tensor([1]*images.shape[0] ).float().cuda())
for i in range(images.s... | device = torch.device("cuda" if torch.cuda.is_available() else "cpu" ) | Natural Language Processing with Disaster Tweets |
14,958,171 | def format_prediction_string(boxes, scores):
pred_strings = []
for j in zip(scores, boxes):
pred_strings.append("{0:.4f} {1} {2} {3} {4}".format(j[0], j[1][0], j[1][1], j[1][2], j[1][3]))
return " ".join(pred_strings )<categorify> | train_df = pd.read_csv('/kaggle/input/nlp-getting-started/train.csv')
train_df.head() | Natural Language Processing with Disaster Tweets |
14,958,171 | results = []
for images, image_ids in data_loader:
predictions = make_tta_predictions(images)
for i, image in enumerate(images):
boxes, scores, labels = run_wbf(predictions, image_index=i)
boxes = boxes.round().astype(np.int32 ).clip(min=0, max=1023)
image_id = image_ids[i]
boxes[:, 2] = boxes[:, 2] - boxes[:, 0]
bo... | print(train_df.info())
print(" | Natural Language Processing with Disaster Tweets |
14,958,171 | test_df = pd.DataFrame(results, columns=['image_id', 'PredictionString'])
test_df.to_csv('submission.csv', index=False)
test_df.head()<install_modules> | train_df['target'].value_counts(normalize=True ) | Natural Language Processing with Disaster Tweets |
14,958,171 | !pip install --no-deps '.. /input/timm-package/timm-0.1.26-py3-none-any.whl' > /dev/null
!pip install --no-deps '.. /input/pycocotools/pycocotools-2.0-cp37-cp37m-linux_x86_64.whl' > /dev/null<set_options> | test_df = pd.read_csv('/kaggle/input/nlp-getting-started/test.csv')
test_df.head() | Natural Language Processing with Disaster Tweets |
14,958,171 | SEED = 42
def seed_everything(seed):
random.seed(seed)
os.environ['PYTHONHASHSEED'] = str(seed)
np.random.seed(seed)
torch.manual_seed(seed)
torch.cuda.manual_seed(seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = True
seed_everything(SEED )<define_variables> | train_df[~train_df['keyword'].isnull() ][['keyword', 'text']] | Natural Language Processing with Disaster Tweets |
14,958,171 | TRAIN_DATA_PATH = '.. /input/global-wheat-detection/train/'
TRAIN_CSV_PATH = '.. /input/global-wheat-detection/train.csv'
TEST_DATA_PATH = '.. /input/global-wheat-detection/test/'
if len(os.listdir('.. /input/global-wheat-detection/test/')) >11:
PL_OPT = True
else:
PL_OPT = False
warmup_opt = True
warmup_epoch = 1
PL_l... | len(train_df['location'].unique() ) | Natural Language Processing with Disaster Tweets |
14,958,171 | def get_train_transforms() :
return A.Compose(
[
A.RandomSizedCrop(min_max_height=(800, 800), height=1024, width=1024, p=0.5),
A.OneOf([
A.HueSaturationValue(hue_shift_limit=0.2, sat_shift_limit= 0.2,
val_shift_limit=0.2, p=0.9),
A.RandomBrightnessContrast(brightness_limit=0.2,
contrast_limit=0.2, p=0.9),
],p=0.9),
A.... | train_df.drop(columns=['keyword', 'location'], inplace=True ) | Natural Language Processing with Disaster Tweets |
14,958,171 | marking = pd.read_csv(TRAIN_CSV_PATH)
bboxs = np.stack(marking['bbox'].apply(lambda x: np.fromstring(x[1:-1], sep=',')))
for i, column in enumerate(['x', 'y', 'w', 'h']):
marking[column] = bboxs[:,i]
marking.drop(columns=['bbox'], inplace=True )<data_type_conversions> | spacy_en = spacy.load('en_core_web_sm', disable=['parser','ner'])
bert_tokenizer = BertTokenizer.from_pretrained('bert-base-uncased', do_lower_case=True ) | Natural Language Processing with Disaster Tweets |
14,958,171 | class DatasetT(Dataset):
def __init__(self, image_ids, transforms=None):
super().__init__()
self.image_ids = image_ids
self.transforms = transforms
def __getitem__(self, index: int):
image_id = self.image_ids[index]
image = cv2.imread(f'{TEST_DATA_PATH}/{image_id}.jpg', cv2.IMREAD_COLOR)
image = cv2.cvtColor(image, cv... | abbreviations = {
"$" : " dollar ",
"€" : " euro ",
"4ao" : "for adults only",
"a.m" : "before midday",
"a3" : "anytime anywhere anyplace",
"aamof" : "as a matter of fact",
"acct" : "account",
"adih" : "another day in hell",
"afaic" : "as far as i am concerned",
"afaict" : "as far as i can tell",
"afaik" : "as far as i... | Natural Language Processing with Disaster Tweets |
14,958,171 | class DatasetRetriever(Dataset):
def __init__(self, marking, image_ids, transforms=None, test=False):
super().__init__()
self.image_ids = image_ids
self.marking = marking
self.transforms = transforms
self.test = test
def __getitem__(self, index: int):
image_id = self.image_ids[index]
if self.test or random.random() > 0... | special_characters = {
"Surṳ":"Suruc",
"JapÌ_n":"Japan" ,
"\x89ÛÏWhen":"When",
"å£3million":"3 million",
"fromåÊwounds":"from wounds",
"m̼sica":"music",
"donå«t":"do not",
"didn`t":"did not",
"i\x89Ûªm":"I am",
"I\x89Ûªm":"I am",
"it\x89Ûªs":"it is",
"It\x89Ûªs":"It is",
"i\x89Ûªd":"I would",
"I\x89Ûªd":"I would",
"... | Natural Language Processing with Disaster Tweets |
14,958,171 | def load_test_net(checkpoint_path):
config = get_efficientdet_config('tf_efficientdet_d5')
net = EfficientDet(config, pretrained_backbone=False)
config.num_classes = 1
config.image_size=img_size
net.class_net = HeadNet(config, num_outputs=config.num_classes, norm_kwargs=dict(eps=.001, momentum=.01))
checkpoint = torc... | expand_contractions = {
"I'm":"I am",
"I'M":"I am",
"i'm":"I am",
"i'M":"I am",
"i'd":"I would",
"I'd":"I would",
"i'll":"I will",
"I'll":"I will",
"i've":"I have",
"I've":"I have",
"you're":"you are",
"You're":"You are",
"you'd":"you would",
"You'd":"You would",
"you've":"you have",
"You've":"You have",
"you'll":"you ... | Natural Language Processing with Disaster Tweets |
14,958,171 | def load_train_net(checkpoint_path):
config = get_efficientdet_config('tf_efficientdet_d5')
net = EfficientDet(config, pretrained_backbone=False)
config.num_classes = 1
config.image_size = img_size
net.class_net = HeadNet(config, num_outputs=config.num_classes, norm_kwargs=dict(eps=.001, momentum=.01))
checkpoint = t... | informal_abbreviations = {
"b/c":"because",
"w/e":"whatever",
"w/out":"without",
"w/o":"without",
"w/":"with ",
"<3":"love",
"c/o":"care of",
"p/u":"pick up",
"
":" "
} | Natural Language Processing with Disaster Tweets |
14,958,171 | class BaseWheatTTA:
image_size = img_size
def augment(self, image):
raise NotImplementedError
def batch_augment(self, images):
raise NotImplementedError
def deaugment_boxes(self, boxes):
raise NotImplementedError
class TTAHorizontalFlip(BaseWheatTTA):
def augment(self, image):
return image.flip(1)
def batch_augmen... | def clean_text(text):
cleaned_text = text.lower()
cleaned_text = re.sub(r'https?:\S+|www\.\S+', '', cleaned_text)
cleaned_text = re.sub(r'<.*?>', '', cleaned_text)
cleaned_text = ''.join(ch for ch in cleaned_text if ch in string.printable)
cleaned_text = ' '.join(abbreviations[word] if word in abbreviations else wor... | Natural Language Processing with Disaster Tweets |
14,958,171 | tta_transforms = []
for tta_combination in product([TTAHorizontalFlip() , None],
[TTAVerticalFlip() , None],
[TTARotate90() , None]):
tta_transforms.append(TTACompose([tta_transform for tta_transform in tta_combination if tta_transform]))<find_best_model_class> | def add_special_token(text):
cleaned_text = "[CLS] " + text + " [SEP]"
return cleaned_text | Natural Language Processing with Disaster Tweets |
14,958,171 | def make_tta_predictions(images, models, score_threshold):
with torch.no_grad() :
images = torch.stack(images ).float().cuda()
assert images.shape[0] == 1
predictions = []
boxes_All = []
scores_All = []
for tta_transform in tta_transforms:
for net in models:
det = net(tta_transform.batch_augment(images.clone()), torch.... | train_df['cleaned_text'] = np.vectorize(clean_text )(train_df['text'])
train_df['cleaned_text'] = np.vectorize(add_special_token )(train_df['cleaned_text'] ) | Natural Language Processing with Disaster Tweets |
14,958,171 | class TrainGlobalConfig:
num_workers = 2
batch_size = PL_batchsize
n_epochs = PL_epoch
lr = PL_lr
folder = 'plabel_model'
verbose = True
verbose_step = 1
step_scheduler = False
validation_scheduler = True
if PL_lr_sche == 'cos':
SchedulerClass = torch.optim.lr_scheduler.CosineAnnealingLR
scheduler_params = dict(
T_max... | MAX_LEN = 128
BATCH_SZ=32 | Natural Language Processing with Disaster Tweets |
14,958,171 | class WarmUp(_LRScheduler):
def __init__(self, optimizer, total_iters, last_epoch=-1):
self.total_iters = total_iters
super(WarmUp, self ).__init__(optimizer, last_epoch)
def get_lr(self):
return [base_lr * self.last_epoch /(self.total_iters + 1e-8)for base_lr in self.base_lrs]<init_hyperparams> | def generate_input_attention_mask(tweets):
tokenized_tweets = [bert_tokenizer.tokenize(tweet)for tweet in tweets]
input_ids = [bert_tokenizer.convert_tokens_to_ids(x)for x in tokenized_tweets]
input_ids = pad_sequences(input_ids, maxlen=MAX_LEN, dtype="long", truncating="post", padding="post")
attention_masks = []
for... | Natural Language Processing with Disaster Tweets |
14,958,171 | warnings.filterwarnings("ignore")
class Fitter:
def __init__(self, model, device, config, train_loader_length):
self.config = config
self.epoch = 0
self.base_dir = f'./{config.folder}'
if not os.path.exists(self.base_dir):
os.makedirs(self.base_dir)
self.log_path = f'{self.base_dir}/log.txt'
self.best_summary_loss = ... | train_input_ids, train_attention_masks = generate_input_attention_mask(train_df['cleaned_text'] ) | Natural Language Processing with Disaster Tweets |
14,958,171 | class AverageMeter(object):
def __init__(self):
self.reset()
def reset(self):
self.val = 0
self.avg = 0
self.sum = 0
self.count = 0
def update(self, val, n=1):
self.val = val
self.sum += val * n
self.count += n
self.avg = self.sum / self.count<categorify> | def recall(y_true, y_pred):
true_positives = K.sum(K.round(y_true * y_pred))
possible_positives = K.sum(y_true)
recall = true_positives /(possible_positives + K.epsilon())
return recall
def precision(y_true, y_pred):
true_positives = K.sum(K.round(y_true * y_pred))
predicted_positives = K.sum(K.round(y_pred))
precisi... | Natural Language Processing with Disaster Tweets |
14,958,171 | if PL_OPT:
test_models = []
for p in path2:
test_models.append(load_test_net(p))
results_plabel = []
for images, image_ids in data_loader:
predictions = make_tta_predictions(images, test_models, PL_thr)
for i, image in enumerate(images):
assert i == 0
image_id = image_ids[i]
image_ = cv2.imread(f'{TEST_DATA_PATH}/{ima... | train_inputs, validation_inputs, train_labels, validation_labels = train_test_split(train_input_ids, train_df['target'], train_size=0.8, random_state=100)
train_masks, validation_masks, _, _ = train_test_split(train_attention_masks, train_df['target'], train_size=0.8, random_state=100)
train_inputs = torch.tensor(tra... | Natural Language Processing with Disaster Tweets |
14,958,171 | def format_prediction_string(boxes, scores):
pred_strings = []
for j in zip(scores, boxes):
pred_strings.append("{0:.4f} {1} {2} {3} {4}".format(j[0], j[1][0], j[1][1], j[1][2], j[1][3]))
return " ".join(pred_strings )<load_pretrained> | train_data = TensorDataset(train_inputs, train_masks, train_labels)
train_sampler = RandomSampler(train_data)
train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=BATCH_SZ)
validation_data = TensorDataset(validation_inputs, validation_masks, validation_labels)
validation_sampler = SequentialS... | Natural Language Processing with Disaster Tweets |
14,958,171 | if PL_OPT:
final_models = [
load_test_net(f'plabel_model/last-checkpoint1.bin')
]
else:
final_models = []
for p in path1:
final_models.append(load_test_net(p))<predict_on_test> | model = BertForSequenceClassification.from_pretrained('bert-base-uncased', num_labels=2)
model.cuda() | Natural Language Processing with Disaster Tweets |
14,958,171 | results = []
for images, image_ids in data_loader:
predictions = make_tta_predictions(images, final_models, OOF_thr)
for i, image in enumerate(images):
assert i == 0
boxes, scores, labels = run_wbf(predictions, img_size, WBF_iou_thr, WBF_skip_thr)
if img_size == 512:
boxes =(boxes*2 ).astype(np.int32 ).clip(min=0, ma... | param_optimizer = list(model.named_parameters())
no_decay = ['bias', 'gamma', 'beta']
optimizer_grouped_parameters = [
{'params': [p for n, p in param_optimizer if not any(nd in n for nd in no_decay)],
'weight_decay_rate': 0.01},
{'params': [p for n, p in param_optimizer if any(nd in n for nd in no_decay)],
'weight_de... | Natural Language Processing with Disaster Tweets |
14,958,171 | test_df = pd.DataFrame(results, columns=['image_id', 'PredictionString'])
test_df.to_csv('submission.csv', index=False)
test_df.head(10 )<import_modules> | train_loss_set = []
epochs = 4
for _ in range(epochs):
model.train()
tr_loss = 0
nb_tr_examples, nb_tr_steps = 0, 0
for step, batch in enumerate(train_dataloader):
batch = tuple(t.to(device)for t in batch)
b_input_ids, b_input_mask, b_labels = batch
optimizer.zero_grad()
loss = model(b_input_ids, token_type_ids=None, ... | Natural Language Processing with Disaster Tweets |
14,958,171 | from object_detection_utils import show_Nimages<set_options> | model.eval()
eval_accuracy = 0
nb_eval_steps = 0
for batch in validation_dataloader:
batch = tuple(t.to(device)for t in batch)
b_input_ids, b_input_mask, b_labels = batch
with torch.no_grad() :
logits = model(b_input_ids, token_type_ids=None, attention_mask=b_input_mask)
logits = logits.detach().cpu().numpy()
label_i... | Natural Language Processing with Disaster Tweets |
14,958,171 | def seed_everything(seed=42):
random.seed(seed)
os.environ['PYTHONHASHSEED'] = str(seed)
np.random.seed(seed)
torch.manual_seed(seed)
torch.cuda.manual_seed(seed)
torch.backends.cudnn.deterministic = True
<load_pretrained> | test_df.drop(columns=['keyword', 'location'], inplace=True)
test_df['cleaned_text'] = np.vectorize(clean_text )(test_df['text'])
test_df['cleaned_text'] = np.vectorize(add_special_token )(test_df['cleaned_text'])
test_input_ids, test_attention_masks = generate_input_attention_mask(test_df['cleaned_text'] ) | Natural Language Processing with Disaster Tweets |
14,958,171 | BEST_PATHS = ["/kaggle/input/best-models-frcnn/F0_68_nofinetune_clear_best.bin",
"/kaggle/input/best-models-frcnn/F1_68_nofinetune_clear_best.bin",
"/kaggle/input/5fold-68-clear/F2_68_nofinetune_clear_best.bin",
"/kaggle/input/5fold-68-clear/F3_68_nofinetune_clear_best.bin"]
for BEST_PATH in BEST_PATHS:
ckp = torch.loa... | test_inputs = torch.tensor(test_input_ids, dtype=torch.long)
test_attention = torch.tensor(test_attention_masks, dtype=torch.long)
test_data = TensorDataset(test_inputs, test_attention)
test_sampler = SequentialSampler(test_data)
test_dataloader = DataLoader(test_data, sampler=test_sampler, batch_size = BATCH_SZ ) | Natural Language Processing with Disaster Tweets |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.