readme add

README.md

@@ -1,59 +1,10 @@
-# 🦙🇺🇿 Llama 3.1 8B Instruct Uzbek — Fine-tune va Chat
-
-Bu Space O‘zbek tilida ishlovchi **Llama 3.1 8B Instruct** modelini **fine-tune qilish** va **interaktiv chatda test qilish** uchun yaratilgan.
-
-## 🚀 Nimalar qilish mumkin?
-
-✅ O‘z dataset'ingiz asosida modelni Streamlit orqali fine-tune qilish  
-✅ Modelga savollar berib interaktiv javob olish  
-✅ Lora (PEFT) yordamida tez va yengil o‘qitish imkoniyati  
-✅ O‘zbek tilidagi AI model bilan bevosita ishlash  
-
 ---
-
-## 📦 Qanday ishlaydi?
-
-1. `datasets/train.json` faylingizni joylang (`instruction`, `input`, `output` formatida)
-2. Modelni yuklang va fine-tune tugmasini bosing
-3. Chat orqali test qilib ko‘ring
-
+title: "Llama 3.1 8B Uzbek Fine-Tuning"
+emoji: "🤖"
+colorFrom: "#000000"
+colorTo: "#00FF00"
+sdk: streamlit
+sdk_version: "1.12.0"
+app_file: app.py
+pinned: false
 ---
-
-## 💾 Dataset namunasi
-
-```json
-[
-  {
-    "instruction": "O‘zbekistonning poytaxti qaysi shahar?",
-    "input": "",
-    "output": "Toshkent shahri"
-  }
-]
-```
-
----
-
-## 🔧 Texnologiyalar
-
-- 🤗 Transformers
-- 🧠 Llama 3.1 8B Instruct (Uzbek)
-- 🪶 PEFT (LoRA)
-- ⚡ Accelerate, bitsandbytes
-- 🧩 Streamlit UI
-
----
-
-## 📤 Kelajak
-
-Yaqin orada quyidagilar qo‘shiladi:
-
-- Modelni Hugging Face Hub'ga push qilish
-- Ko‘p epoch, live progress bar
-- Tuning parametrlarini sozlash oynasi
-- Dataset yuklashni Streamlit ichida qilish
-
----
-
-**Muallif:** [@qahorovz](https://t.me/qahorovz)  
-**AI kuchi bilan — O‘zbekiston uchun! 🇺🇿**
-