Update app.py

app.py

@@ -14,7 +14,7 @@ warnings.filterwarnings('ignore')
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 print(f"Using device: {device}")
 
-model_name = 'cognitivecomputations/dolphin-vision-72b'
+model_name = 'cognitivecomputations/dolphin-vision-7b'
 
 # create model and load it to the specified device
 model = AutoModelForCausalLM.from_pretrained(
@@ -29,8 +29,9 @@ tokenizer = AutoTokenizer.from_pretrained(
     trust_remote_code=True
 )
 
-def inference(prompt, image, temperature, beam_size):
+def inference(prompt, image, temperature, beam_size, system_instruction):
     messages = [
+        {"role": "system", "content": system_instruction},
         {"role": "user", "content": f'<image>\n{prompt}'}
     ]
     text = tokenizer.apply_chat_template(
@@ -65,6 +66,11 @@ def inference(prompt, image, temperature, beam_size):
 with gr.Blocks() as demo:
     with gr.Row():
         with gr.Column():
+            system_instruction = gr.Textbox(
+                label="System Instruction",
+                value="You are Dolphin, a helpful AI assistant",
+                lines=2
+            )
             prompt_input = gr.Textbox(label="Prompt", placeholder="Describe this image in detail")
             image_input = gr.Image(label="Image", type="pil")
             temperature_input = gr.Slider(minimum=0.1, maximum=2.0, value=0.7, step=0.1, label="Temperature")
@@ -75,7 +81,7 @@ with gr.Blocks() as demo:
 
     submit_button.click(
         fn=inference, 
-        inputs=[prompt_input, image_input, temperature_input, beam_size_input], 
+        inputs=[prompt_input, image_input, temperature_input, beam_size_input, system_instruction], 
         outputs=output_text
     )