Update app.py

app.py

@@ -1,64 +1,93 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+# Initialize Inference Clients for all models
+paligemma224_client = InferenceClient("google/paligemma2-3b-pt-224")
+paligemma448_client = InferenceClient("google/paligemma2-3b-pt-448")
+paligemma896_client = InferenceClient("google/paligemma2-3b-pt-896")
+paligemma28b_client = InferenceClient("google/paligemma2-28b-pt-224")
+llama_client = InferenceClient("llama/3.3-1b")
+deepseek_client = InferenceClient("deepseek-ai/deepseek-vl2")
+omniparser_client = InferenceClient("microsoft/OmniParser")
+pixtral_client = InferenceClient("mistralai/Pixtral-12B-2409")
 
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+def enhance_prompt(prompt: str) -> str:
+    # Use the Paligemma models for prompt enhancement
+    prompt_224 = paligemma224_client.infer(prompt)
+    prompt_448 = paligemma448_client.infer(prompt)
+    prompt_896 = paligemma896_client.infer(prompt)
+    
+    # Combine all enhanced prompts into a single one
+    enhanced_prompt = f"Enhanced (224): {prompt_224}\nEnhanced (448): {prompt_448}\nEnhanced (896): {prompt_896}"
+    
+    # Ultra-enhance the prompt using Paligemma 28b
+    ultra_enhanced_prompt = paligemma28b_client.infer(enhanced_prompt)
+    
+    return ultra_enhanced_prompt
+
+
+def generate_answer(enhanced_prompt: str) -> str:
+    # Generate answers using the three models: llama, deepseek, and omniparser
+    llama_answer = llama_client.infer(enhanced_prompt)
+    deepseek_answer = deepseek_client.infer(enhanced_prompt)
+    omniparser_answer = omniparser_client.infer(enhanced_prompt)
+    
+    # Combine answers from all models
+    combined_answer = f"Llama: {llama_answer}\nDeepseek: {deepseek_answer}\nOmniparser: {omniparser_answer}"
+    
+    return combined_answer
+
+
+def enhance_answer(answer: str) -> str:
+    # Enhance the final answer using Pixtral model
+    enhanced_answer = pixtral_client.infer(answer)
+    return enhanced_answer
+
+
+def process(message: str) -> str:
+    # Step 1: Enhance the prompt using the Paligemma models
+    enhanced_prompt = enhance_prompt(message)
+    
+    # Step 2: Generate an answer using the three models
+    answer = generate_answer(enhanced_prompt)
+    
+    # Step 3: Enhance the generated answer using Pixtral
+    final_answer = enhance_answer(answer)
+    
+    return final_answer
 
+
+# Gradio interface setup
+def respond(message, history: list[tuple[str, str]], system_message, max_tokens, temperature, top_p):
+    # Include system message and history in conversation
+    messages = [{"role": "system", "content": system_message}]
     for val in history:
         if val[0]:
             messages.append({"role": "user", "content": val[0]})
         if val[1]:
             messages.append({"role": "assistant", "content": val[1]})
-
-    messages.append({"role": "user", "content": message})
-
+    
+    # Get the final enhanced response
+    final_answer = process(message)
+    
+    # Yield the response for the Gradio interface
     response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
+    for token in final_answer:
         response += token
         yield response
 
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
+# Gradio interface setup
 demo = gr.ChatInterface(
     respond,
     additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
+        gr.Textbox(value="You are a helpful assistant.", label="System message"),
         gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new 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,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
+        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
     ],
 )
 
-
 if __name__ == "__main__":
     demo.launch()