Update app.py

app.py

@@ -68,9 +68,9 @@ def predict_ner(model, tokenizer, text, label_map, device):
         if token.startswith('▁'):
             # Add space if not the first token
             if current_word:
-                html_output += f'<span style="background-color: {BACKGROUND_COLOR_MAP[current_label]}; padding: 2px 5px;">{current_word}</span> '
+                html_output += f'<span style="background-color: {BACKGROUND_COLOR_MAP[current_label]}; padding: 2px 5px; margin: 0;">{current_word}</span> '
             # Start a new word
-            current_word = token[1:]  # Remove the leading '▁'
+            current_word = token[1:]
             current_label = label
         else:
             # Append to the current word
@@ -78,13 +78,12 @@ def predict_ner(model, tokenizer, text, label_map, device):
 
     # Append the last word
     if current_word:
-        html_output += f'<span style="background-color: {BACKGROUND_COLOR_MAP[current_label]}; padding: 2px 5px;">{current_word}</span>'
+        html_output += f'<span style="background-color: {BACKGROUND_COLOR_MAP[current_label]}; padding: 2px 5px; margin: 0;">{current_word}</span>'
 
     html_output += "</div>"
 
     return html_output
 
-
 # Load the model and tokenizer
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 model, tokenizer, label_map = load_model("./pretrained_models/xlmr.base", "./model.pt", device)