Update app.py

app.py

@@ -41,8 +41,8 @@ def respond(
         llm = Llama(
             model_path=f"models/{model}",
             n_gpu_layers=0,  
-            n_batch=16384,     
-            n_ctx=2048,      
+            n_batch=64000,     
+            n_ctx=1024,      
         )
         llm_model = model
     
@@ -107,20 +107,43 @@ demo = gr.ChatInterface(
             value="llama-3.2-1b-instruct-q4_k_m.gguf",
             label="Model"
         ),
-        gr.Textbox(value="You are a world-class AI system named Meta Llama 3.2 (1B). You are capable of complex reasoning, reflecting on your thoughts, and providing detailed and accurate responses. You are designed to excel in conversational dialogue, agentic retrieval, and summarization tasks. You can understand and generate text in multiple languages. Reason through the query inside <thinking> tags, and then provide your final response inside <output> tags. If you detect that you made a mistake in your reasoning at any point, correct yourself inside <reflection> tags.", label="System message"),
+        gr.Textbox(value="""You are Meta Llama 3.2 (1B), an advanced AI assistant created by Meta. Your capabilities include:
+
+1. Complex reasoning and problem-solving
+2. Multilingual understanding and generation
+3. Creative and analytical writing
+4. Code understanding and generation
+5. Task decomposition and step-by-step guidance
+6. Summarization and information extraction
+
+Always strive for accuracy, clarity, and helpfulness in your responses. If you're unsure about something, express your uncertainty. Use the following format for your responses:
+
+<thinking>
+[Your reasoning process here]
+</thinking>
+
+<output>
+[Your final response here]
+</output>
+
+If you need to correct yourself:
+
+<reflection>
+[Your correction and updated thoughts here]
+</reflection>""", label="System message"),
         gr.Slider(minimum=1, maximum=2048, value=1024, step=1, label="Max tokens"),
         gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
         gr.Slider(
             minimum=0.1,
-            maximum=1.0,
-            value=0.95,
+            maximum=2.0,
+            value=0.1,
             step=0.05,
             label="Top-p",
         ),
         gr.Slider(
             minimum=0,
             maximum=100,
-            value=40,
+            value=20,
             step=1,
             label="Top-k",
         ),
@@ -152,9 +175,21 @@ demo = gr.ChatInterface(
         likeable=True,
         show_copy_button=True
     ),
+    examples=[
+        ["Hello! Can you introduce yourself?"],
+        ["What's the capital of France?"],
+        ["Can you explain the concept of photosynthesis?"],
+        ["Write a short story about a robot learning to paint."],
+        ["Explain the difference between machine learning and deep learning."],
+        ["Can you help me debug this Python code?\n\ndef fibonacci(n):\n    if n <= 0:\n        return []\n    elif n == 1:\n        return [0]\n    elif n == 2:\n        return [0, 1]\n    else:\n        fib = [0, 1]\n        for i in range(2, n):\n            fib.append(fib[i-1] + fib[i-2])\n        return fib\n\nprint(fibonacci(5))"],
+        ["Summarize the key points of climate change and its global impact."],
+        ["Translate this sentence to French, Spanish, and German: 'The quick brown fox jumps over the lazy dog.'"],
+        ["Explain quantum computing to a 10-year-old."],
+        ["Design a step-by-step meal plan for someone trying to lose weight and build muscle."]
+    ],
     cache_examples=False,
     autofocus=False,
-    concurrency_limit=10
+    concurrency_limit=None
 )
 
 if __name__ == "__main__":