Update app.py

app.py

@@ -1,3 +1,42 @@
 import gradio as gr
+from transformers import AutoModelForSequenceClassification, AutoTokenizer
 
-gr.Interface.load("models/tkurtulus/rottentomato-classifier").launch()
+def classify_sentence(sent:str):
+    toksentence = tokenizer(sent,truncation=True,return_tensors="pt")
+    model.eval()
+    with torch.no_grad():
+        toksentence.to(device)
+        output = model(**toksentence)
+    
+    return F.softmax(output.logits,dim=1).argmax(dim=1)
+
+    
+def classify_text(text:str):
+    sentences = sent_tokenize(text)
+    annotations = np.array(list(map(classify_sentence,sentences)),dtype=object)  
+    result = list(zip(sentences,[mapping[val] for val in annotations]))
+    return (annotations,result)
+    
+def classify_text_wrapper(text:str):
+    preds,result = classify_text(text)
+    n = len(preds)
+    non_biased = np.where(preds==0)[0].shape[0]
+    biased = np.where(preds==1)[0].shape[0]
+    
+    return (result,{'bias ratio':biased/n})
+
+examples=[["[Newsoms's] obsession with masks has created an almost hostile environment in our neighborhoods and streets.\n“He won because the Election was Rigged,” Trump wrote, not referring to Biden by name, adding a list of complaints about vote counting"]]
+
+model = AutoModelForSequenceClassification.from_pretrained("tkurtulus/autotrain-rottentomato-2981285985")
+tokenizer = AutoTokenizer.from_pretrained("tkurtulus/autotrain-rottentomato-2981285985");
+model.eval();
+
+label = gr.outputs.Label(num_top_classes=None,label='')
+text_h = gr.outputs.HighlightedText(color_map={'Unbiased':'#9ad1A1','Biased':'#db8a8a'},label='Classification')
+inputs = gr.inputs.Textbox(placeholder=None, default="", label=None)
+
+app = gr.Interface(fn=classify_text_wrapper,title='Bias classifier',theme='default',
+                    inputs="textbox",layout='unaligned', outputs=[text_h,label], capture_session=True
+                    ,examples=examples)
+
+app.launch(inbrowser=True)