Update app.py

app.py

@@ -1,19 +1,14 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
-from typing import List, Dict, Any, Tuple  # ← + Tuple!
+from typing import List, Dict, Any, Tuple
 
-# 10 моделей
+# ТОП-5 MODELS (chat-ready, быстрые, HF-tested)
 MODELS = [
-    "microsoft/Phi-3-mini-4k-instruct",
-    "Qwen/Qwen2.5-0.5B-Instruct",
-    "Qwen/Qwen2.5-1.5B-Instruct",
-    "google/gemma-2-2b-it",
-    "HuggingFaceH4/zephyr-7b-beta",
-    "mistralai/Mistral-7B-Instruct-v0.3",
-    "mistralai/Mistral-Nemo-Instruct-2407",
-    "microsoft/Phi-3.5-mini-instruct",
-    "NousResearch/Hermes-2-Theta-Llama-3.1-8B",
-    "cognitivecomputations/dolphin-2.9-llama3-8b"
+    "Qwen/Qwen2.5-0.5B-Instruct",     # ⚡ Супер-быстрая (0.5B)
+    "Qwen/Qwen2.5-1.5B-Instruct",     # Быстрая (1.5B)
+    "microsoft/Phi-3-mini-4k-instruct", # Надёжная Phi-3
+    "mistralai/Mistral-7B-Instruct-v0.3", # Классика
+    "HuggingFaceH4/zephyr-7b-beta"    # Zephyr (chat-tuned)
 ]
 
 def respond(message: str, 
@@ -21,8 +16,10 @@ def respond(message: str,
             model_id: str, 
             system_prompt: str,
             hf_token: str) -> Tuple[List[Dict[str, str]], str, Dict[str, Any]]:
-    """HF API с native messages."""
+    """HF API с debug."""
     try:
+        print(f"🚀 Запрос: Model={model_id}, Token={'Yes' if hf_token else 'No'}, Msg='{message[:50]}...'")  # Лог HF
+        
         client = InferenceClient(model=model_id, token=hf_token.strip() or None)
         
         messages = []
@@ -39,6 +36,7 @@ def respond(message: str,
         )
         
         bot_reply = response.choices[0].message.content
+        print(f"✅ Ответ: {bot_reply[:50]}...")  # Лог успеха
         
         new_history = history + [
             {"role": "user", "content": message},
@@ -47,51 +45,48 @@ def respond(message: str,
         return new_history, "", gr.update(value="")
     
     except Exception as e:
-        error_msg = f"❌ {str(e)}"
+        full_error = f"❌ Model: {model_id}\nError: {str(e)}\n"
+        print(f"💥 ERROR: {full_error}")  # Лог HF
+        
         if "429" in str(e) or "rate" in str(e).lower():
-            error_msg += "\n🔥 Rate limit → HF Token!"
-        elif "token" in str(e).lower():
-            error_msg += "\n🔑 Проверь токен."
+            full_error += "🔥 RATE LIMIT! Вставь HF Token (huggingface.co/settings/tokens)."
+        elif "token" in str(e).lower() or "Unauthorized" in str(e):
+            full_error += "🔑 Неверный/нет Token. Создай новый."
+        elif "No chat template" in str(e) or "tokenizer" in str(e):
+            full_error += "🧠 Модель не chat-ready. Выбери другую (Qwen/Phi)."
+        else:
+            full_error += "🌐 HF API глючит. Попробуй Token или другую модель."
         
         new_history = history + [
             {"role": "user", "content": message},
-            {"role": "assistant", "content": error_msg}
+            {"role": "assistant", "content": full_error}
         ]
-        return new_history, error_msg, gr.update(value="")
+        return new_history, full_error, gr.update(value="")
 
 # UI
-with gr.Blocks(title="🚀 HF Чат-Тестер (готово!)", theme=gr.themes.Soft()) as demo:
-    gr.Markdown("# Тест HF Inference API\n**Token** → 1M tokens/мес с тебя. [Токен](https://huggingface.co/settings/tokens)")
+with gr.Blocks(title="🚀 HF Чат (debug!)", theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# Тест HF API\n**1. Выбери Qwen2.5-0.5B** ⚡\n**2. Вставь Token** → Нет лимитов\n[Token](https://huggingface.co/settings/tokens)")
     
     with gr.Row(variant="compact"):
         model_dropdown = gr.Dropdown(choices=MODELS, value=MODELS[0], label="🧠 Модель")
-        system_prompt = gr.Textbox(label="📝 System", placeholder="Ты ИИ-хакер.", lines=2)
+        system_prompt = gr.Textbox(label="📝 System", placeholder="Ты ИИ-ученик.", lines=2)
         hf_token = gr.Textbox(label="🔑 Token", placeholder="hf_...", type="password")
     
     chatbot = gr.Chatbot(type="messages", height=500)
     
     with gr.Row():
-        msg_input = gr.Textbox(placeholder="Сообщение... (Enter)", scale=7, container=True)
+        msg_input = gr.Textbox(placeholder="Привет! (Enter)", scale=7)
         send_btn = gr.Button("📤", variant="primary", scale=1)
     
     with gr.Row():
-        clear_btn = gr.Button("🗑️ Очистить")
-        retry_btn = gr.Button("🔄 Повторить")
+        clear_btn = gr.Button("🗑️ Clear")
+        retry_btn = gr.Button("🔄 Retry")
     
-    status = gr.Textbox(label="Статус", interactive=False)
+    status = gr.Textbox(label="Логи/Статус", interactive=False, lines=3)
     
     # Events
-    send_btn.click(
-        fn=respond,
-        inputs=[msg_input, chatbot, model_dropdown, system_prompt, hf_token],
-        outputs=[chatbot, status, msg_input]
-    )
-    
-    msg_input.submit(
-        fn=respond,
-        inputs=[msg_input, chatbot, model_dropdown, system_prompt, hf_token],
-        outputs=[chatbot, status, msg_input]
-    )
+    send_btn.click(fn=respond, inputs=[msg_input, chatbot, model_dropdown, system_prompt, hf_token], outputs=[chatbot, status, msg_input])
+    msg_input.submit(fn=respond, inputs=[msg_input, chatbot, model_dropdown, system_prompt, hf_token], outputs=[chatbot, status, msg_input])
     
     def clear():
         return [], "", gr.update(value="")