Update app.py

app.py

@@ -6,7 +6,6 @@ import spaces
 import threading
 
 # --- 1. Model and Processor Setup ---
-
 model_id = "bharatgenai/patram-7b-instruct"
 device = "cuda" if torch.cuda.is_available() else "cpu"
 print(f"Using device: {device}")
@@ -21,6 +20,11 @@ model = AutoModelForCausalLM.from_pretrained(
 )
 print("Model and processor loaded successfully.")
 
+# Default system prompt
+DEFAULT_SYSTEM_PROMPT = """You are Patram, a helpful AI assistant created by BharatGenAI. You are designed to analyze images and answer questions about them.
+Think step by step before providing your answers. Be detailed, accurate, and helpful in your responses.
+You can understand both text and image inputs to provide comprehensive answers to user queries."""
+
 # --- Define and apply a more flexible chat template ---
 chat_template = """{% for message in messages %}
     {{ message['role'].capitalize() }}: {{ message['content'] }}
@@ -49,8 +53,11 @@ def generate_response(user_message, messages_list, image_pil, max_new_tokens, to
             add_generation_prompt=True
         )
 
-        # Preprocess image and the entire formatted prompt
-        inputs = processor.process(images=[image_pil], text=prompt)
+        if image_pil:
+            # Preprocess image and the entire formatted prompt
+            inputs = processor.process(images=[image_pil], text=prompt)
+        else: 
+            inputs = processor.process(text=prompt)
         inputs = {k: v.to(device).unsqueeze(0) for k, v in inputs.items()}
         inputs = {k: v.half() if v.dtype == torch.float32 else v for k, v in inputs.items()}
 
@@ -93,9 +100,6 @@ def process_chat(user_message, chatbot_display, messages_list, image_pil, max_ne
     """
     This function handles the chat logic for a single turn with streaming.
     """
-    if image_pil is None:
-        chatbot_display.append((user_message, "Please upload an image first to start the conversation."))
-        return chatbot_display, messages_list, ""
 
     # Append user's message to the chatbot display list
     chatbot_display.append((user_message, ""))
@@ -113,7 +117,7 @@ def process_chat(user_message, chatbot_display, messages_list, image_pil, max_ne
 
 def clear_chat():
     """Resets the chat, history, and image."""
-    return [], [], None, "", 256, 0.9, 50, 0.6
+    return [], [], None, "", 256, 0.9, 50, 0.6, DEFAULT_SYSTEM_PROMPT
 
 # --- 3. Gradio Interface Definition ---
 with gr.Blocks(theme=gr.themes.Default(primary_hue="blue", secondary_hue="neutral")) as demo:
@@ -127,6 +131,12 @@ with gr.Blocks(theme=gr.themes.Default(primary_hue="blue", secondary_hue="neutra
     with gr.Row():
         with gr.Column(scale=1):
             image_input_render = gr.Image(type="pil", label="Upload Image")
+            system_prompt = gr.Textbox(
+                label="System Prompt",
+                value=DEFAULT_SYSTEM_PROMPT,
+                interactive=True,
+                lines=5
+            )
             clear_btn = gr.Button("🗑️ Clear Chat and Image")
             with gr.Accordion("Generation Parameters", open=False):
                 max_new_tokens = gr.Slider(minimum=32, maximum=4096, value=256, step=32, label="Max New Tokens")
@@ -149,6 +159,20 @@ with gr.Blocks(theme=gr.themes.Default(primary_hue="blue", secondary_hue="neutra
                 )
                 submit_btn = gr.Button("Send", variant="primary", scale=1, min_width=0)
 
+    # Initialize messages_list with system prompt
+    demo.load(
+        fn=lambda: [{"role": "system", "content": DEFAULT_SYSTEM_PROMPT}],
+        inputs=None,
+        outputs=messages_list
+    )
+
+    # Update messages_list when system prompt changes
+    system_prompt.change(
+        fn=lambda system_prompt: [{"role": "system", "content": system_prompt}],
+        inputs=system_prompt,
+        outputs=messages_list
+    )
+
     # --- Event Listeners ---
 
     # Define the action for submitting a message (via button or enter key)
@@ -167,7 +191,7 @@ with gr.Blocks(theme=gr.themes.Default(primary_hue="blue", secondary_hue="neutra
     clear_btn.click(
         fn=clear_chat,
         inputs=[],
-        outputs=[chatbot_display, messages_list, image_input_render, user_textbox, max_new_tokens, top_p, top_k, temperature],
+        outputs=[chatbot_display, messages_list, image_input_render, user_textbox, max_new_tokens, top_p, top_k, temperature, system_prompt],
         queue=False
     )