add app

app.py

@@ -33,7 +33,7 @@ def predict(model_name, text):
     tokenizer = MODEL_BUF["tokenizer"]
     model = MODEL_BUF["model"]
 
-    unsmile_labels = ["여성/가족","남성","성소수자","인종/국적","연령","지역","종교","기타 혐오","악플/욕설","clean", 'name', 'number', 'address', 'bank', 'person']
+    unsmile_labels = ["여성/가족","남성","성소수자","인종/국적","연령","지역","종교","기타 혐오","악플/욕설", "clean", '이름', '전화번호', '주소', '계좌번호', '주민번호']
     num_labels = len(unsmile_labels)
 
     model.config.id2label = {i: label for i, label in zip(range(num_labels), unsmile_labels)}
@@ -49,7 +49,9 @@ def predict(model_name, text):
     return pipe(text)[0]
 
 if __name__ == '__main__':
-    text = '읿딴걸 홍볿글 읿랉곭 쌑젩낄고 앉앟있냩'
+    exam1 = '경기도 성남시 수정구 태평3동은 우리 동네야!'
+    exam2 = '내 핸드폰 번호는 010-3930-8237 이야!'
+    exam3 = '아 젠장 너무 짜증난다'
 
     model_name_list = [
         'momo/KcELECTRA-base_Hate_speech_Privacy_Detection',
@@ -60,77 +62,12 @@ if __name__ == '__main__':
     app = gr.Interface(
         fn=predict,
         inputs=[gr.inputs.Dropdown(model_name_list, label="Model Name"), 'text'], outputs='text', 
-        examples = [[MODEL_BUF["name"], text], [MODEL_BUF["name"], "4=🦀 4≠🦀"]],
+        examples = [
+            [MODEL_BUF["name"], exam1], 
+            [MODEL_BUF["name"], exam2],
+            [MODEL_BUF["name"], exam3]
+            ],
         title="한국어 혐오표현, 개인정보 판별기 (Korean Hate Speech and Privacy Detection)",
-        description="Korean Hate Speech and Privacy Detection."
+        description="Korean Hate Speech and Privacy Detection. \t 15개 label Detection: 여성/가족, 남성, 성소수자, 인종/국적, 연령, 지역, 종교, 기타 혐오, 악플/욕설, clean, name, number, address, bank, person"
         )
-    app.launch()
-    
-
-# # global var
-# MODEL_NAME = 'jason9693/SoongsilBERT-base-beep'
-# tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
-# model = AutoModelForSequenceClassification.from_pretrained(MODEL_NAME)
-# config = AutoConfig.from_pretrained(MODEL_NAME)
-
-# MODEL_BUF = {
-#     "name": MODEL_NAME,
-#     "tokenizer": tokenizer,
-#     "model": model,
-#     "config": config
-# }
-
-# def change_model_name(name):
-#     MODEL_BUF["name"] = name
-#     MODEL_BUF["tokenizer"] = AutoTokenizer.from_pretrained(name)
-#     MODEL_BUF["model"] = AutoModelForSequenceClassification.from_pretrained(name)
-#     MODEL_BUF["config"] = AutoConfig.from_pretrained(name)
-
-
-# def predict(model_name, text):
-#     if model_name != MODEL_BUF["name"]:
-#         change_model_name(model_name)
-    
-#     tokenizer = MODEL_BUF["tokenizer"]
-#     model = MODEL_BUF["model"]
-#     config = MODEL_BUF["config"]
-
-#     tokenized_text = tokenizer([text], return_tensors='pt')
-
-#     input_tokens = tokenizer.convert_ids_to_tokens(tokenized_text.input_ids[0])
-#     try:
-#         input_tokens = util.bytetokens_to_unicdode(input_tokens) if config.model_type in ['roberta', 'gpt', 'gpt2'] else input_tokens
-#     except KeyError:
-#         input_tokens = input_tokens
-
-#     model.eval()
-#     output, attention = model(**tokenized_text, output_attentions=True, return_dict=False)
-#     output = F.softmax(output, dim=-1)
-#     result = {}
-    
-#     for idx, label in enumerate(output[0].detach().numpy()):
-#         result[config.id2label[idx]] = float(label)
-
-#     fig = visualize_attention(input_tokens, attention[0][0].detach().numpy())
-#     return result, fig#.logits.detach()#.numpy()#, output.attentions.detach().numpy()
-
-
-# if __name__ == '__main__':
-#     text = '읿딴걸 홍볿글 읿랉곭 쌑젩낄고 앉앟있냩'
-
-#     model_name_list = [
-#         'jason9693/SoongsilBERT-base-beep',
-#         "beomi/beep-klue-roberta-base-hate",
-#         "beomi/beep-koelectra-base-v3-discriminator-hate",
-#         "beomi/beep-KcELECTRA-base-hate"
-#     ]
-
-#     #Create a gradio app with a button that calls predict()
-#     app = gr.Interface(
-#         fn=predict,
-#         inputs=[gr.inputs.Dropdown(model_name_list, label="Model Name"), 'text'], outputs=['label', 'plot'], 
-#         examples = [[MODEL_BUF["name"], text], [MODEL_BUF["name"], "4=🦀 4≠🦀"]],
-#         title="한국어 혐오성 발화 분류기 (Korean Hate Speech Classifier)",
-#         description="Korean Hate Speech Classifier with Several Pretrained LM\nCurrent Supported Model:\n1. SoongsilBERT\n2. KcBERT(+KLUE)\n3. KcELECTRA\n4.KoELECTRA."
-#         )
-#     app.launch(inline=False)
+    app.launch()