{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 🚀 VIBE CODER v2.0 MAX — Official Google Colab Studio\n",
    "\n",
    "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/shawaz03/LLM/blob/main/vibe_coder_quickstart.ipynb)\n",
    "\n",
    "Run **Vibe Coder v2.0 MAX (7B)** on a **Free Google Colab T4 GPU** in 4-bit NF4 Quantization (5.2 GB VRAM, 0 memory errors).\n",
    "\n",
    "- **Hugging Face Model**: [`shawaz03/vibe-coder-7b-max`](https://huggingface.co/shawaz03/vibe-coder-7b-max)\n",
    "- **Specialization**: Autonomous Full-Stack React 19, Next.js 15, TypeScript, Tailwind CSS, Prisma, and Anti-AI Aesthetic Directives."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1. Install Dependencies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "!pip install -q transformers torch accelerate bitsandbytes huggingface_hub"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. Load Model in 4-bit (Fast & Zero Memory Overhead)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig, TextStreamer\n",
    "\n",
    "MODEL_ID = \"shawaz03/vibe-coder-7b-max\"\n",
    "print(f\"🚀 Loading {MODEL_ID} in 4-bit NF4...\")\n",
    "\n",
    "bnb_config = BitsAndBytesConfig(\n",
    "    load_in_4bit=True,\n",
    "    bnb_4bit_quant_type=\"nf4\",\n",
    "    bnb_4bit_use_double_quant=True,\n",
    "    bnb_4bit_compute_dtype=torch.float16,\n",
    ")\n",
    "\n",
    "tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, trust_remote_code=True)\n",
    "model = AutoModelForCausalLM.from_pretrained(\n",
    "    MODEL_ID,\n",
    "    quantization_config=bnb_config,\n",
    "    device_map=\"auto\",\n",
    "    trust_remote_code=True,\n",
    ")\n",
    "model.eval()\n",
    "streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)\n",
    "\n",
    "print(\"\\n✅ VIBE CODER IS READY!\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3. Vibe Coder Engine Definition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "VIBE_CODER_SYSTEM_PROMPT = \"\"\"You are Vibe Coder, a world-class principal full-stack software engineer and UI/UX designer.\n",
    "ENGINEERING & CODING DIRECTIVES:\n",
    "1. COMPLETE IMPLEMENTATIONS: Never output '// TODO', '/* implement later */', or incomplete stubs. Every component, hook, and route must be fully written.\n",
    "2. NO ARTIFACT STRINGS: Never output template tags, variant numbers (e.g. 'Build Variant #...'), or internal directive texts.\n",
    "3. FULL STATE & INTERACTION: Include real mock data arrays, working toggle logic, and complete TypeScript types.\n",
    "4. MODERN DESIGN SYSTEM: Use Tailwind CSS with dark neutral palettes (bg-neutral-900, border-neutral-800), clean accents (cyan, emerald, violet), and Lucide React icons.\n",
    "5. CLEAN OUTPUT: Output standard, clean TypeScript/React code with 'use client' when state is used.\"\"\"\n",
    "\n",
    "def ask_vibe_coder(prompt_text, temperature=0.2, max_tokens=1500):\n",
    "    if not prompt_text.strip():\n",
    "        print(\"⚠️ Please enter a prompt in the box below.\")\n",
    "        return\n",
    "\n",
    "    messages = [\n",
    "        {\"role\": \"system\", \"content\": VIBE_CODER_SYSTEM_PROMPT},\n",
    "        {\"role\": \"user\", \"content\": prompt_text}\n",
    "    ]\n",
    "    \n",
    "    formatted_prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)\n",
    "    inputs = tokenizer(formatted_prompt, return_tensors=\"pt\").to(\"cuda\")\n",
    "    \n",
    "    print(\"=\" * 75)\n",
    "    print(f\"💬 USER PROMPT: {prompt_text}\")\n",
    "    print(\"=\" * 75 + \"\\n\")\n",
    "    \n",
    "    with torch.no_grad():\n",
    "        model.generate(\n",
    "            **inputs,\n",
    "            streamer=streamer,\n",
    "            max_new_tokens=max_tokens,\n",
    "            temperature=temperature,\n",
    "            top_p=0.95,\n",
    "            repetition_penalty=1.05,\n",
    "            do_sample=True,\n",
    "            eos_token_id=tokenizer.eos_token_id,\n",
    "            pad_token_id=tokenizer.pad_token_id,\n",
    "        )\n",
    "    print(\"\\n\" + \"=\" * 75)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4. 🎮 Dynamic Prompt Studio (Interactive Form Box)\n",
    "Type your custom prompt into the box below and click the **Play (Run)** button:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "cellView": "form"
   },
   "outputs": [],
   "source": [
    "# @title 🚀 Vibe Coder Interactive Code Generator { vertical-output: true }\n",
    "# @markdown Enter your custom prompt and configure generation settings:\n",
    "\n",
    "prompt = \"Build an interactive pricing matrix in React with Tailwind CSS, supporting monthly/annual billing toggle, 5 feature bullet checkmarks, and popular badge with zero placeholders.\"  # @param {type:\"string\"}\n",
    "temperature = 0.2  # @param {type:\"slider\", min:0.05, max:1.0, step:0.05}\n",
    "max_tokens = 1500  # @param {type:\"slider\", min:256, max:3072, step:128}\n",
    "\n",
    "ask_vibe_coder(prompt, temperature=temperature, max_tokens=max_tokens)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5. 💬 Continuous Live Chat Loop\n",
    "Run this cell to chat and prompt continuously in real-time:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "while True:\n",
    "    user_query = input(\"\\n👉 Type any prompt (or 'exit' to quit): \")\n",
    "    if not user_query.strip() or user_query.lower() in [\"exit\", \"quit\"]:\n",
    "        print(\"Session stopped.\")\n",
    "        break\n",
    "    ask_vibe_coder(user_query, temperature=0.2)"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "gpuType": "T4",
   "provenance": []
  },
  "language_info": {
   "name": "python"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}