Increase context to 1024 and use larger example prompt.

app.py

@@ -33,7 +33,7 @@ def generate_text(prompt, example):
 
     logging.info(f"Input : {input}")
 
-    output_stream = llm(input, max_tokens=200, stop=["</s>", "###"], stream=True)
+    output_stream = llm(input, max_tokens=1024, stop=["</s>", "###"], stream=True)
 
     full_text = ""
     for output_chunk in output_stream:
@@ -44,14 +44,14 @@ def generate_text(prompt, example):
     logging.info(f"Generated text: {full_text}")
 
 model = download_model()
-llm = Llama(model_path=model)
+llm = Llama(model_path=model, n_ctx=1024)
 
 # Create a Gradio interface
 interface = gr.Interface(
   fn=generate_text,
   inputs=[
       gr.Textbox(lines=4, placeholder="Enter text to anonimize...", label="Text with PII",
-      value="My name is Filipe and my phone number is 555-121-2234. How are you?")
+      value="My name is Filipe and my phone number is 555-121-2234. How are you?\nWant to meet up in Los Angeles at 5pm by the Grove?\nI live in downtown LA.")
   ],
   outputs=gr.Textbox(label="PII Sanitized version of the text"),
   title="PII Sanitization Model",