def test_classify_text():
happy_tc = HappyTextClassification(
model_type="DISTILBERT",
model_name="distilbert-base-uncased-finetuned-sst-2-english")
result = happy_tc.classify_text("What a great movie")
assert result.label == 'POSITIVE'
assert result.score > 0.9
def example_2_0():
happy_tc_distilbert = HappyTextClassification(
) # default with "distilbert-base-uncased"
happy_tc_albert = HappyTextClassification(model_type="ALBERT",
model_name="albert-base-v2")
happy_tc_bert = HappyTextClassification("BERT", "bert-base-uncased")
happy_tc_roberta = HappyTextClassification("ROBERTA",
"deepset/roberta-base-squad2")
def example_2_0():
happy_tc_distilbert = HappyTextClassification("DISTILBERT",
"distilbert-base-uncased",
num_labels=2) # default
happy_tc_albert = HappyTextClassification(model_type="ALBERT",
model_name="albert-base-v2")
happy_tc_bert = HappyTextClassification("BERT", "bert-base-uncased")
happy_tc_roberta = HappyTextClassification("ROBERTA", "roberta-base")
def example_2_3():
happy_tc = HappyTextClassification(
model_type="DISTILBERT",
model_name="distilbert-base-uncased-finetuned-sst-2-english",
num_labels=2) # Don't forget to set num_labels!
result = happy_tc.eval("../../data/tc/train-eval.csv")
print(type(result)) # <class 'happytransformer.happy_trainer.EvalResult'>
print(result) # EvalResult(eval_loss=0.007262040860950947)
print(result.loss) # 0.007262040860950947
def example_2_3():
happy_tc = HappyTextClassification(
model_type="DISTILBERT",
model_name="distilbert-base-uncased-finetuned-sst-2-english",
num_labels=2) # Don't forget to set num_labels!
args = TCEvalArgs(
save_preprocessed_data=False) # for demonstration -- not needed
result = happy_tc.eval("../../data/tc/train-eval.csv", args=args)
print(type(result)) # <class 'happytransformer.happy_trainer.EvalResult'>
print(result) # EvalResult(eval_loss=0.007262040860950947)
print(result.loss) # 0.007262040860950947
def example_2_1():
happy_tc = HappyTextClassification(
model_type="DISTILBERT",
model_name="distilbert-base-uncased-finetuned-sst-2-english")
result = happy_tc.classify_text("Great movie! 5/5")
print(
type(result)
) # <class 'happytransformer.happy_text_classification.TextClassificationResult'>
print(
result
) # TextClassificationResult(label='LABEL_1', score=0.9998761415481567)
print(result.label) # LABEL_1
def test_tc_test():
happy_tc = HappyTextClassification(
model_type="DISTILBERT",
model_name="distilbert-base-uncased-finetuned-sst-2-english")
result = happy_tc.test("../data/tc/test.csv")
answer = [
TextClassificationResult(label='POSITIVE', score=0.9998401999473572),
TextClassificationResult(label='NEGATIVE', score=0.9772131443023682),
TextClassificationResult(label='NEGATIVE', score=0.9966067671775818),
TextClassificationResult(label='POSITIVE', score=0.9792295098304749)
]
assert result == answer
def test_tc_save():
happy = HappyTextClassification()
happy.save("model/")
result_before = happy.classify_text("What a great movie")
happy = HappyTextClassification(load_path="model/")
result_after = happy.classify_text("What a great movie")
assert result_before.label == result_after.label
def example_2_4():
happy_tc = HappyTextClassification(
model_type="DISTILBERT",
model_name="distilbert-base-uncased-finetuned-sst-2-english",
num_labels=2) # Don't forget to set num_labels!
result = happy_tc.test("../../data/tc/test.csv")
print(type(result)) # <class 'list'>
print(
result
) # [TextClassificationResult(label='LABEL_1', score=0.9998401999473572), TextClassificationResult(label='LABEL_0', score=0.9772131443023682)...
print(
type(result[0])
) # <class 'happytransformer.happy_text_classification.TextClassificationResult'>
print(
result[0]
) # TextClassificationResult(label='LABEL_1', score=0.9998401999473572)
print(result[0].label) # LABEL_1
def test_tc_train_effectiveness():
"""assert that training decreases the loss"""
happy_tc = HappyTextClassification(model_type="DISTILBERT",
model_name="distilbert-base-uncased")
before_loss = happy_tc.eval("../data/tc/train-eval.csv").loss
happy_tc.train("../data/tc/train-eval.csv")
after_loss = happy_tc.eval("../data/tc/train-eval.csv").loss
assert after_loss < before_loss
def test_tc_train_effectiveness_multi():
happy_tc = HappyTextClassification(model_type="DISTILBERT",
model_name="distilbert-base-uncased",
num_labels=3)
before_loss = happy_tc.eval("../data/tc/train-eval-multi.csv").loss
happy_tc.train("../data/tc/train-eval-multi.csv")
after_loss = happy_tc.eval("../data/tc/train-eval-multi.csv").loss
assert after_loss < before_loss
def example_2_5():
happy_tc = HappyTextClassification(
model_type="DISTILBERT",
model_name="distilbert-base-uncased-finetuned-sst-2-english",
num_labels=2) # Don't forget to set num_labels!
before_loss = happy_tc.eval("../../data/tc/train-eval.csv").loss
happy_tc.train("../../data/tc/train-eval.csv")
after_loss = happy_tc.eval("../../data/tc/train-eval.csv").loss
print("Before loss: ", before_loss) # 0.007262040860950947
print("After loss: ", after_loss) # 0.000162081079906784
def test_tc_with_dataclass():
happy_tc = HappyTextClassification()
train_args = TCTrainArgs(learning_rate=0.01, num_train_epochs=1)
happy_tc.train("../data/tc/train-eval.csv", args=train_args)
eval_args = TCEvalArgs()
result_eval = happy_tc.eval("../data/tc/train-eval.csv", args=eval_args)
test_args = TCTestArgs()
result_test = happy_tc.test("../data/tc/test.csv", args=test_args)
def test_tc_with_dic():
happy_tc = HappyTextClassification()
train_args = {'learning_rate': 0.01, "num_train_epochs": 1}
happy_tc.train("../data/tc/train-eval.csv", args=train_args)
eval_args = {}
result_eval = happy_tc.eval("../data/tc/train-eval.csv", args=eval_args)
test_args = {}
result_test = happy_tc.test("../data/tc/test.csv", args=test_args)
def test_tc_eval():
happy_tc = HappyTextClassification(
model_type="DISTILBERT",
model_name="distilbert-base-uncased-finetuned-sst-2-english")
results = happy_tc.eval("../data/tc/train-eval.csv")
assert results.loss == approx(0.007262040860950947, 0.01)
def test_tc_train():
happy_tc = HappyTextClassification(
model_type="DISTILBERT",
model_name="distilbert-base-uncased-finetuned-sst-2-english")
results = happy_tc.train("../data/tc/train-eval.csv")
def test_tc_save_load_eval():
happy_wp = HappyTextClassification()
output_path = "data/tc-train.json"
data_path = "../data/tc/train-eval.csv"
run_save_load(happy_wp, output_path, ARGS_TC_EVAL, data_path, "eval")
from webapp.models import Report
from webapp import api
from webapp import tweepy
import pandas as pd
import preprocessor as p
from happytransformer import HappyTextClassification
happy_tc = HappyTextClassification(
'BERT', 'Hate-speech-CNERG/dehatebert-mono-english', 2)
model = happy_tc
def scan(username):
tweets_d = tweetpull(username) # get the target's tweets
total_tweets = len(tweets_d) # gets length of total tweets
tweets = [] # buffer for tweets
counter = -1 # counter to track id
for tweet in tweets_d['tweet_text']: # iterates through tweets
counter += 1
tweets_d['prediction'] = model.classify_text(
tweet) # sets prediction value
if 'LABEL_1' in str(tweets_d['prediction']): # if hate
id = tweets_d['tweet_id'][counter] # gets tweet id
tweets.append([id, tweet])
total_scanned_tweets = len(tweets) # amount of hate tweets
account_summary = get_account_summary(
username, total_scanned_tweets,
total_tweets) # makes call to account summary
profile = get_twitter_info(username)
return tweets, account_summary, profile # returns results (bad tweet ids, the tweet text and the overall account
# summary) and account summary and profile
def example_2_2():
happy_tc = HappyTextClassification(
model_type="DISTILBERT",
model_name="distilbert-base-uncased-finetuned-sst-2-english",
num_labels=2) # Don't forget to set num_labels!
happy_tc.train("../../data/tc/train-eval.csv")