Update app.py

app.py

@@ -1,132 +1,83 @@
 import gradio as gr
-import requests
-import os
-import json
+from transformers import pipeline
+import torch
 
 # ===============================
-# 🔧 Settings
+# 🤖 Load Model LOCALLY
 # ===============================
-MODEL_NAME = os.environ.get("MODEL_NAME", "kawkabelaloom/astramind")
-HF_TOKEN = os.environ.get("HF_TOKEN", "")
+MODEL_NAME = "kawkabelaloom/astramind"
 
-# ⬅️ الحل هنا
-API_URL = f"https://api-inference.huggingface.co/models/{MODEL_NAME}?task=text-generation"
-
-headers = {
-    "Authorization": f"Bearer {HF_TOKEN}",
-    "Content-Type": "application/json"
-} if HF_TOKEN else {}
-
-# ===============================
-# 🤖 Chatbot
-# ===============================
-class AstramindChatbot:
-    def __init__(self):
-        self.history = []
-        self.system_prompt = (
-            "أنت مساعد عربي ذكي اسمك أسترا. "
-            "أجب بلغة عربية واضحة ومفيدة.\n\n"
-        )
-
-    def build_prompt(self, user_message):
-        prompt = self.system_prompt
-        for msg in self.history[-6:]:
-            if msg["role"] == "user":
-                prompt += f"المستخدم: {msg['content']}\n"
-            else:
-                prompt += f"المساعد: {msg['content']}\n"
-        prompt += f"المستخدم: {user_message}\nالمساعد:"
-        return prompt
-
-    def send_to_model(self, user_message, max_tokens, temperature):
-        try:
-            prompt = self.build_prompt(user_message)
-
-            payload = {
-                "inputs": prompt,
-                "parameters": {
-                    "max_new_tokens": max_tokens,
-                    "temperature": temperature,
-                    "top_p": 0.9,
-                    "do_sample": True,
-                    "return_full_text": False
-                }
-            }
-
-            print("\n📤 PROMPT:\n", prompt)
-            print("\n📤 PAYLOAD:\n", json.dumps(payload, ensure_ascii=False, indent=2))
-
-            response = requests.post(
-                API_URL,
-                headers=headers,
-                json=payload,
-                timeout=60
-            )
-
-            print("📥 STATUS:", response.status_code)
-            print("📥 RAW:", response.text[:500])
-
-            if response.status_code == 200:
-                result = response.json()
-
-                if isinstance(result, list) and "generated_text" in result[0]:
-                    answer = result[0]["generated_text"]
-                else:
-                    answer = str(result)
-
-                self.history.append({"role": "user", "content": user_message})
-                self.history.append({"role": "assistant", "content": answer})
-
-                return answer.strip()
-
-            elif response.status_code == 503:
-                return "🔄 النموذج بيحمّل… حاول بعد دقيقة"
-
-            else:
-                return f"❌ API ERROR {response.status_code}\n{response.text}"
-
-        except Exception as e:
-            return f"❌ PYTHON ERROR\n{str(e)}"
-
-    def chat(self, msg, max_tokens, temp):
-        if not msg.strip():
-            return "⚠️ اكتب رسالة"
-        return self.send_to_model(msg, max_tokens, temp)
+generator = pipeline(
+    "text-generation",
+    model=MODEL_NAME,
+    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
+    device=0 if torch.cuda.is_available() else -1
+)
 
 # ===============================
-# 🚀 Init
+# 🧠 Chat Logic
 # ===============================
-bot = AstramindChatbot()
+history = []
+
+SYSTEM_PROMPT = (
+    "أنت مساعد عربي ذكي اسمك أسترا. "
+    "أجب بلغة عربية واضحة ومفيدة.\n\n"
+)
+
+def build_prompt(user_message):
+    prompt = SYSTEM_PROMPT
+    for h in history[-6:]:
+        if h["role"] == "user":
+            prompt += f"المستخدم: {h['content']}\n"
+        else:
+            prompt += f"المساعد: {h['content']}\n"
+    prompt += f"المستخدم: {user_message}\nالمساعد:"
+    return prompt
+
+def chat(user_message, chat_history, max_tokens, temperature):
+    if not user_message.strip():
+        return chat_history, ""
+
+    prompt = build_prompt(user_message)
+
+    output = generator(
+        prompt,
+        max_new_tokens=max_tokens,
+        temperature=temperature,
+        do_sample=True,
+        return_full_text=False
+    )[0]["generated_text"]
+
+    history.append({"role": "user", "content": user_message})
+    history.append({"role": "assistant", "content": output})
+
+    chat_history.append((user_message, output))
+    return chat_history, ""
+
+def clear_chat():
+    history.clear()
+    return []
 
 # ===============================
-# 🎛️ UI
+# 🖥️ Gradio UI
 # ===============================
-def respond(msg, history, tokens, temp):
-    history = history or []
-    reply = bot.chat(msg, tokens, temp)
-    history.append((msg, reply))
-    return history, "", "✅ تم التنفيذ"
-
-def clear():
-    bot.history = []
-    return [], "🗑️ تم المسح"
-
 with gr.Blocks(title="Astramind") as demo:
     gr.Markdown("# 🤖 Astramind Chatbot")
+    gr.Markdown("تشغيل محلي داخل HuggingFace Space")
 
-    status = gr.Textbox("جاهز", interactive=False)
-    chat_ui = gr.Chatbot(height=400)
+    chatbot = gr.Chatbot(height=400)
 
     with gr.Row():
         msg = gr.Textbox(placeholder="اكتب رسالتك...", lines=2)
         send = gr.Button("إرسال")
 
     with gr.Accordion("⚙️ الإعدادات"):
-        tokens = gr.Slider(50, 500, 200, step=10)
-        temp = gr.Slider(0.1, 1.0, 0.7)
+        tokens = gr.Slider(50, 500, 200, step=10, label="طول الرد")
+        temp = gr.Slider(0.1, 1.0, 0.7, step=0.1, label="الإبداع")
+
+    send.click(chat, [msg, chatbot, tokens, temp], [chatbot, msg])
+    msg.submit(chat, [msg, chatbot, tokens, temp], [chatbot, msg])
 
-    send.click(respond, [msg, chat_ui, tokens, temp], [chat_ui, msg, status])
-    msg.submit(respond, [msg, chat_ui, tokens, temp], [chat_ui, msg, status])
-    gr.Button("مسح").click(clear, outputs=[chat_ui, status])
+    gr.Button("مسح المحادثة").click(clear_chat, outputs=chatbot)
 
-demo.launch(server_name="0.0.0.0", server_port=7860, debug=True)
+demo.launch()