Update app.py

app.py

@@ -116,7 +116,7 @@ def predict(text):
 
 with gr.Blocks() as demo:
     gr.Markdown(
-        """
+        """\
         ## Detect text generated using LLMs 🤖
 
         Linguistic features such as Perplexity and other SOTA methods such as GLTR were used to classify between Human written and LLM Generated \
@@ -124,31 +124,36 @@ with gr.Blocks() as demo:
 
         - Source & Credits: [https://github.com/Hello-SimpleAI/chatgpt-comparison-detection](https://github.com/Hello-SimpleAI/chatgpt-comparison-detection)
         - Competition: [https://www.kaggle.com/competitions/llm-detect-ai-generated-text/leaderboard](https://www.kaggle.com/competitions/llm-detect-ai-generated-text/leaderboard)
-        - Solution WriteUp: [https://www.kaggle.com/competitions/llm-detect-ai-generated-text/discussion/470224](https://www.kaggle.com/competitions/llm-detect-ai-generated-text/discussion/470224)
-
-        ### Linguistic Analysis: Language Model Perplexity
-        The perplexity (PPL) is commonly used as a metric for evaluating the performance of language models (LM). It is defined as the exponential \
-        of the negative average log-likelihood of the text under the LM. A lower PPL indicates that the language model is more confident in its \
-        predictions, and is therefore considered to be a better model. The training of LMs is carried out on large-scale text corpora, it can \
-        be considered that it has learned some common language patterns and text structures. Therefore, PPL can be used to measure how \
-        well a text conforms to common characteristics.
-
-        ### GLTR: Giant Language Model Test Room
-        This idea originates from the following paper: arxiv.org/pdf/1906.04043.pdf. It studies 3 tests to compute features of an input text. Their \
-        major assumption is that to generate fluent and natural-looking text, most decoding strategies sample high probability tokens from the head \
-        of the distribution. I selected the most powerful Test-2 feature, which is the number of tokens in the Top-10, Top-100, Top-1000, and 1000+ \
-        ranks from the LM predicted probability distributions.
-
-        ### Modelling
-        Scikit-learn's VotingClassifier consisting of XGBClassifier, LGBMClassifier, CatBoostClassifier and RandomForestClassifier with default parameters
+        - Solution WriteUp: [https://www.kaggle.com/competitions/llm-detect-ai-generated-text/discussion/470224](https://www.kaggle.com/competitions/llm-detect-ai-generated-text/discussion/470224)\
         """
     )
-    a1 = gr.Textbox( lines=7, label='Text', value=example )
-    button1 = gr.Button("🤖 Predict!")
-    gr.Markdown("Prediction:")
-    label1 = gr.Textbox(lines=1, label='Predicted Label')
-    score1 = gr.Textbox(lines=1, label='Predicted Probability')
-
-    button1.click(predict, inputs=[a1], outputs=[label1, score1])
+    with gr.Column():
+        gr.Markdown(
+            """\
+            ### Linguistic Analysis: Language Model Perplexity
+            The perplexity (PPL) is commonly used as a metric for evaluating the performance of language models (LM). It is defined as the exponential \
+            of the negative average log-likelihood of the text under the LM. A lower PPL indicates that the language model is more confident in its \
+            predictions, and is therefore considered to be a better model. The training of LMs is carried out on large-scale text corpora, it can \
+            be considered that it has learned some common language patterns and text structures. Therefore, PPL can be used to measure how \
+            well a text conforms to common characteristics.
+    
+            ### GLTR: Giant Language Model Test Room
+            This idea originates from the following paper: arxiv.org/pdf/1906.04043.pdf. It studies 3 tests to compute features of an input text. Their \
+            major assumption is that to generate fluent and natural-looking text, most decoding strategies sample high probability tokens from the head \
+            of the distribution. I selected the most powerful Test-2 feature, which is the number of tokens in the Top-10, Top-100, Top-1000, and 1000+ \
+            ranks from the LM predicted probability distributions.
+    
+            ### Modelling
+            Scikit-learn's VotingClassifier consisting of XGBClassifier, LGBMClassifier, CatBoostClassifier and RandomForestClassifier with default parameters\
+            """
+        )
+        with gr.Group()
+            a1 = gr.Textbox( lines=7, label='Text', value=example )
+            button1 = gr.Button("🤖 Predict!")
+            gr.Markdown("Prediction:")
+            label1 = gr.Textbox(lines=1, label='Predicted Label')
+            score1 = gr.Textbox(lines=1, label='Predicted Probability')
+        
+            button1.click(predict, inputs=[a1], outputs=[label1, score1])
 
 demo.launch()