Spaces:
Sleeping
Sleeping
Update game1.py
Browse files
game1.py
CHANGED
|
@@ -114,7 +114,8 @@ def func1(lang_selected, num_selected, human_predict, num1, num2, user_important
|
|
| 114 |
# Use a pipeline as a high-level helper
|
| 115 |
from transformers import pipeline
|
| 116 |
|
| 117 |
-
|
|
|
|
| 118 |
output = classifier([text['text']])
|
| 119 |
|
| 120 |
star2num = {
|
|
@@ -322,7 +323,8 @@ def func1_written(text_written, human_predict, lang_written):
|
|
| 322 |
# tokenizer = AutoTokenizer.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment")
|
| 323 |
# model = AutoModelForSequenceClassification.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment")
|
| 324 |
|
| 325 |
-
|
|
|
|
| 326 |
|
| 327 |
output = classifier([text_written])
|
| 328 |
|
|
@@ -353,10 +355,12 @@ def func1_written(text_written, human_predict, lang_written):
|
|
| 353 |
import shap
|
| 354 |
|
| 355 |
# sentiment_classifier = pipeline("text-classification", return_all_scores=True)
|
|
|
|
|
|
|
| 356 |
if lang_written == "Dutch":
|
| 357 |
-
sentiment_classifier = pipeline("text-classification", model='DTAI-KULeuven/robbert-v2-dutch-sentiment', return_all_scores=True)
|
| 358 |
else:
|
| 359 |
-
sentiment_classifier = pipeline("text-classification", model='distilbert-base-uncased-finetuned-sst-2-english', return_all_scores=True)
|
| 360 |
|
| 361 |
explainer = shap.Explainer(sentiment_classifier)
|
| 362 |
|
|
|
|
| 114 |
# Use a pipeline as a high-level helper
|
| 115 |
from transformers import pipeline
|
| 116 |
|
| 117 |
+
device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
|
| 118 |
+
classifier = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer, device=device)
|
| 119 |
output = classifier([text['text']])
|
| 120 |
|
| 121 |
star2num = {
|
|
|
|
| 323 |
# tokenizer = AutoTokenizer.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment")
|
| 324 |
# model = AutoModelForSequenceClassification.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment")
|
| 325 |
|
| 326 |
+
device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
|
| 327 |
+
classifier = pipeline("sentiment-analysis", model="nlptown/bert-base-multilingual-uncased-sentiment", device=device)
|
| 328 |
|
| 329 |
output = classifier([text_written])
|
| 330 |
|
|
|
|
| 355 |
import shap
|
| 356 |
|
| 357 |
# sentiment_classifier = pipeline("text-classification", return_all_scores=True)
|
| 358 |
+
device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
|
| 359 |
+
|
| 360 |
if lang_written == "Dutch":
|
| 361 |
+
sentiment_classifier = pipeline("text-classification", model='DTAI-KULeuven/robbert-v2-dutch-sentiment', return_all_scores=True, device=device)
|
| 362 |
else:
|
| 363 |
+
sentiment_classifier = pipeline("text-classification", model='distilbert-base-uncased-finetuned-sst-2-english', return_all_scores=True, device=device)
|
| 364 |
|
| 365 |
explainer = shap.Explainer(sentiment_classifier)
|
| 366 |
|