use smaller repetition penalty; add doc

app.py

@@ -42,7 +42,7 @@ if __name__ == "__main__":
         grammar_processor = GrammarConstrainedLogitsProcessor(grammar)
 
         outputs = model.generate(
-            **inputs, max_new_tokens=50, repetition_penalty=1.1, return_dict_in_generate=True, output_scores=True, logits_processor=[grammar_processor]
+            **inputs, max_new_tokens=50, repetition_penalty=1.05, return_dict_in_generate=True, output_scores=True, logits_processor=[grammar_processor]
         )
         # Important: don't forget to set `normalize_logits=True` to obtain normalized probabilities (i.e. sum(p) = 1)
         transition_scores = model.compute_transition_scores(outputs.sequences, outputs.scores, normalize_logits=True)
@@ -70,19 +70,19 @@ if __name__ == "__main__":
     with demo:
         gr.Markdown(
             """
-            # 🌈 Color Coded Text Generation 🌈
-            This is a demo of how you can obtain the probabilities of each generated token, and use them to
-            color code the model output.
-            Feel free to clone this demo and modify it to your needs 🤗
-            Internally, it relies on [`compute_transition_scores`](https://huggingface.co/docs/transformers/main/en/main_classes/text_generation#transformers.GenerationMixin.compute_transition_scores),
-            which was added in `transformers` v4.26.0.
+            # Grammar-Constrained Decoding with GPT-2 
+            This is a demo of how you can constrain the output of a GPT-2 model using a formal grammar.
+            Here we use a simple JSON grammar to constrain the output of the model to be valid JSON strings.
+            The grammar is defined in `json_minimal.ebnf` and is written in the Extended Backus-Naur Form (EBNF).
+            Internally, it relies on the library [`transformers-cfg`](https://github.com/epfl-dlab/transformers-CFG).
+            For demo purpose, gpt2 is used, but you can use much larger models for better performance.
             """
         )
 
         with gr.Row():
             with gr.Column():
                 prompt = gr.Textbox(label="Prompt", lines=3, value="This is a valid json string for http request:")
-                button = gr.Button(f"Generate with {MODEL_NAME}, using sampling!")
+                button = gr.Button(f"Generate with json object using {MODEL_NAME}!")
             with gr.Column():
                 highlighted_text = gr.HighlightedText(
                     label="Highlighted generation",