Update app.py

app.py

@@ -17,7 +17,7 @@ model_id = "llava-hf/llava-interleave-qwen-0.5b-hf"
 
 processor = LlavaProcessor.from_pretrained(model_id)
 
-model = LlavaForConditionalGeneration.from_pretrained(model_id, low_cpu_mem_usage=True)
+model = LlavaForConditionalGeneration.from_pretrained(model_id)
 model.to("cpu")
 
 
@@ -82,7 +82,6 @@ client_llama = InferenceClient("meta-llama/Meta-Llama-3-8B-Instruct")
 # Define the main chat function
 def respond(message, history):
     func_caller = []
-    vqa = ""
 
     user_prompt = message
     # Handle image processing
@@ -107,7 +106,7 @@ def respond(message, history):
         ]
     
         message_text = message["text"]
-        func_caller.append({"role": "user", "content": f'[SYSTEM]You are a helpful assistant. You have access to the following functions: \n {str(functions_metadata)}\n\nTo use these functions respond with:\n<functioncall> {{ "name": "function_name", "arguments": {{ "arg_1": "value_1", "arg_1": "value_1", ... }} }}  </functioncall>  [USER] {message_text} {vqa}'})
+        func_caller.append({"role": "user", "content": f'[SYSTEM]You are a helpful assistant. You have access to the following functions: \n {str(functions_metadata)}\n\nTo use these functions respond with:\n<functioncall> {{ "name": "function_name", "arguments": {{ "arg_1": "value_1", "arg_1": "value_1", ... }} }}  </functioncall>  [USER] {message_text}'})
     
         response = client_gemma.chat_completion(func_caller, max_tokens=150)
         response = str(response)
@@ -134,7 +133,7 @@ def respond(message, history):
                 for msg in history:
                     messages += f"\n<|im_start|>user\n{str(msg[0])}<|im_end|>"
                     messages += f"\n<|im_start|>assistant\n{str(msg[1])}<|im_end|>"
-                messages+=f"\n<|im_start|>user\n{message_text} {vqa}<|im_end|>\n<|im_start|>web_result\n{web2}<|im_end|>\n<|im_start|>assistant\n"
+                messages+=f"\n<|im_start|>user\n{message_text}<|im_end|>\n<|im_start|>web_result\n{web2}<|im_end|>\n<|im_start|>assistant\n"
                 stream = client_mixtral.text_generation(messages, max_new_tokens=2000, do_sample=True, stream=True, details=True, return_full_text=False)
                 output = ""
                 for response in stream:
@@ -146,7 +145,7 @@ def respond(message, history):
                 gr.Info("Generating Image, Please wait 10 sec...")
                 seed = random.randint(1, 99999)
                 query = query.replace(" ", "%20")
-                image = f"![](https://image.pollinations.ai/prompt/{query}?seed={seed})"
+                image = f"![](https://image.pollinations.ai/prompt/{message_text}{query}?seed={seed})"
                 yield image
                 time.sleep(8)
                 gr.Info("We are going to Update Our Image Generation Engine to more powerful ones in Next Update. ThankYou")
@@ -167,7 +166,7 @@ def respond(message, history):
                 for msg in history:
                     messages += f"\n<|start_header_id|>user\n{str(msg[0])}<|end_header_id|>"
                     messages += f"\n<|start_header_id|>assistant\n{str(msg[1])}<|end_header_id|>"
-                messages+=f"\n<|start_header_id|>user\n{message_text} {vqa}<|end_header_id|>\n<|start_header_id|>assistant\n"
+                messages+=f"\n<|start_header_id|>user\n{message_text}<|end_header_id|>\n<|start_header_id|>assistant\n"
                 stream = client_llama.text_generation(messages, max_new_tokens=2000, do_sample=True, stream=True, details=True, return_full_text=False)
                 output = ""
                 for response in stream:
@@ -179,7 +178,7 @@ def respond(message, history):
             for msg in history:
                 messages += f"\n<|start_header_id|>user\n{str(msg[0])}<|end_header_id|>"
                 messages += f"\n<|start_header_id|>assistant\n{str(msg[1])}<|end_header_id|>"
-            messages+=f"\n<|start_header_id|>user\n{message_text} {vqa}<|end_header_id|>\n<|start_header_id|>assistant\n"
+            messages+=f"\n<|start_header_id|>user\n{message_text}<|end_header_id|>\n<|start_header_id|>assistant\n"
             stream = client_llama.text_generation(messages, max_new_tokens=2000, do_sample=True, stream=True, details=True, return_full_text=False)
             output = ""
             for response in stream: