Add top-tier models: Llama, Qwen, DeepSeek, Mistral, Gemma

app.py

@@ -1,100 +1,109 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
-import os
 
-# Initialize the Inference Client
-client = InferenceClient(token=os.environ.get("HF_TOKEN"))
-
-# List of available models (you can add more)
+# List of top-tier models from various providers available on Hugging Face
 AVAILABLE_MODELS = [
-    "meta-llama/Llama-3.3-70B-Instruct",
+    # Meta Llama models
+    "meta-llama/Meta-Llama-3.1-70B-Instruct",
+    "meta-llama/Meta-Llama-3.1-8B-Instruct",
+    
+    # Qwen models (Alibaba)
+    "Qwen/Qwen2.5-72B-Instruct",
+    "Qwen/Qwen2.5-7B-Instruct",
     "Qwen/Qwen2.5-Coder-32B-Instruct",
+    
+    # DeepSeek models
+    "deepseek-ai/DeepSeek-V3",
+    "deepseek-ai/DeepSeek-R1-Distill-Qwen-32B",
+    
+    # Mistral models
+    "mistralai/Mixtral-8x7B-Instruct-v0.1",
     "mistralai/Mistral-7B-Instruct-v0.3",
-    "google/gemma-2-9b-it",
+    
+    # Microsoft models
     "microsoft/Phi-3.5-mini-instruct",
+    
+    # Google models
+    "google/gemma-2-27b-it",
+    "google/gemma-2-9b-it",
+    
+    # Other top models
     "HuggingFaceH4/zephyr-7b-beta",
-    "tiiuae/falcon-7b-instruct",
     "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO",
 ]
 
 def chat_with_models(message, history, selected_models):
-    """
-    Send message to selected models and return their responses
-    """
     if not selected_models:
-        return history + [(message, "⚠️ Please select at least one model first!")]
-    
+        return history + [(message, "⚠️ Please select at least one model!")]
     if len(selected_models) > 5:
-        return history + [(message, "⚠️ Please select maximum 5 models only!")]
+        return history + [(message, "⚠️ Maximum 5 models allowed!")]
     
-    # Create response text
     responses = []
-    
     for model_name in selected_models:
         try:
-            response = client.chat_completion(
-                model=model_name,
-                messages=[{"role": "user", "content": message}],
-                max_tokens=500,
-            )
-            model_response = response.choices[0].message.content
-            responses.append(f"**{model_name}:**\n{model_response}\n")
+            client = InferenceClient(model=model_name)
+            try:
+                result = client.chat_completion(
+                    messages=[{"role": "user", "content": message}],
+                    max_tokens=500
+                )
+                response = result.choices[0].message.content
+            except:
+                result = client.text_generation(message, max_new_tokens=300)
+                response = result
+            responses.append(f"**{model_name}:**\n{response}\n")
         except Exception as e:
-            responses.append(f"**{model_name}:**\n❌ Error: {str(e)}\n")
-    
-    combined_response = "\n---\n\n".join(responses)
-    return history + [(message, combined_response)]
+            error = str(e)
+            if "loading" in error.lower():
+                responses.append(f"**{model_name}:**\n⏳ Loading...\n")
+            else:
+                responses.append(f"**{model_name}:**\n❌ {error[:100]}\n")
+    return history + [(message, "\n---\n\n".join(responses))]
 
-# Create the Gradio interface
-with gr.Blocks(title="Anki-Chat: Multi-Model Chat", theme=gr.themes.Soft()) as demo:
+with gr.Blocks(title="Anki-Chat", theme=gr.themes.Soft()) as demo:
     gr.Markdown(
-        """
-        # 🤖 Anki-Chat: Multi-Model Chat Interface
-        
-        Select up to 5 AI models and chat with them simultaneously to compare their responses!
-        """
+        "# 🤖 Anki-Chat: Multi-Model Comparison\n"
+        "Chat with top AI models: Meta Llama, Qwen, DeepSeek, Mistral, Google & more!"
     )
-    
     with gr.Row():
         with gr.Column(scale=1):
             gr.Markdown("### Select Models (Max 5)")
-            model_checkboxes = gr.CheckboxGroup(
+            checkboxes = gr.CheckboxGroup(
                 choices=AVAILABLE_MODELS,
                 label="Available Models",
-                value=[AVAILABLE_MODELS[0]],  # Default to first model
+                value=[AVAILABLE_MODELS[1]]
             )
-        
-        with gr.Column(scale=2):
-            chatbot = gr.Chatbot(
-                label="Chat",
-                height=500,
+            gr.Markdown(
+                "**Providers:**\n"
+                "- 🦙 Meta Llama 3.1\n"
+                "- 💉 Qwen 2.5 (Alibaba)\n"
+                "- 🦉 DeepSeek V3 & R1\n"
+                "- ✨ Mistral\n"
+                "- 🌐 Google Gemma\n"
+                "- 🔥 Microsoft Phi"
             )
-            
+        with gr.Column(scale=2):
+            chatbot = gr.Chatbot(label="Chat", height=500)
             with gr.Row():
                 msg = gr.Textbox(
-                    label="Your Message",
-                    placeholder="Type your message here...",
-                    scale=4,
+                    label="Message",
+                    placeholder="Type here...",
+                    scale=4
                 )
-                send_btn = gr.Button("Send", scale=1, variant="primary")
-            
-            clear_btn = gr.Button("Clear Chat")
+                send = gr.Button("Send", scale=1, variant="primary")
+            clear = gr.Button("Clear")
     
-    # Event handlers
-    msg.submit(chat_with_models, [msg, chatbot, model_checkboxes], [chatbot]).then(
+    msg.submit(chat_with_models, [msg, chatbot, checkboxes], [chatbot]).then(
         lambda: "", None, [msg]
     )
-    send_btn.click(chat_with_models, [msg, chatbot, model_checkboxes], [chatbot]).then(
+    send.click(chat_with_models, [msg, chatbot, checkboxes], [chatbot]).then(
         lambda: "", None, [msg]
     )
-    clear_btn.click(lambda: [], None, [chatbot])
+    clear.click(lambda: [], None, [chatbot])
     
     gr.Markdown(
-        """
-        ---
-        **Note:** This app uses Hugging Face's Inference API. Some models may take longer to respond or may be rate-limited.
-        """
+        "**Note:** Using HF Serverless Inference API. "
+        "Large models may take 20-30s to load initially."
     )
 
-if __name__ == "__main__":
-    demo.launch()
+demo.launch()