alpacaQ&A.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "gpuClass": "standard"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "# Alpaca-Q/A\n",
        "\n",
        "This notebook contains minimal code for running within a UI created by Pynecone [Alpaca](https://github.com/tloen/alpaca-lora/) for demonstration purposes. Please check the repo for more details."
      ],
      "metadata": {
        "id": "yPahm2jZ-RNI"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## NoteS\n",
        "- Remember this 7-billion fined-tuned model on data generated by GPT 3, it's small model compared to chat gpt, running is small gpu, it will not provide the best answer, and it will take time to load - it was created for education purpose only - https://github.com/tloen/alpaca-lora/ \n",
        "- Building the app will take around 6 min or may be more depends on your connection (download rate, that google gives you)\n",
        "- it's slow compared to chat-gpt, but functional, it's just slow\n",
        "- follow up by having `logs.txt` opened, so you can understand what's going on, refresh `logs.txt` to see updates, by reopening it\n",
        "- Web app url is kind of buggy, you might need to refresh\n",
        "- honstly it's hard to keep tracking of what's going, just run all the cells and refresh the url or rerun the `APP-Cell` \n",
        "- ignore warnings\n",
        "- better test within colab cells, you might rarely need to rerun the app-cell\n",
        "- these lines indicates that your app is ready to take off \n",
        "<!-- ```bash\n",
        "wait  - compiling \n",
        "event - compiled client and server succesfully in xxxx ms \n",
        "``` -->"
      ],
      "metadata": {
        "id": "FjP3ohnML2Bw"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "![download.png](data:image/png;base64,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)"
      ],
      "metadata": {
        "id": "lzkJnfc3NkcP"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Keep this tab alive to prevent Colab from disconnecting you { display-mode: \"form\" }\n",
        "\n",
        "#@markdown Press play on the music player that will appear below:\n",
        "%%html\n",
        "<audio src=\"https://oobabooga.github.io/silence.m4a\" controls>"
      ],
      "metadata": {
        "id": "OYWrYOoWEGeH",
        "outputId": "0bcf2069-91dc-4aef-f27e-f7a07ecac0c0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 75
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              "<audio src=\"https://oobabooga.github.io/silence.m4a\" controls>\n"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@title  dependencies  \n",
        "from IPython.display import clear_output\n",
        "%load_ext autoreload\n",
        "%autoreload 2\n",
        "\n",
        "clear_output()\n",
        "!pip install -q bitsandbytes\n",
        "\n",
        "# alpaca training dataset - training LLAMA is byond being done on Tesla T4, well at least i don't know how to do quantized training yet\n",
        "!pip install -q datasets loralib sentencepiece \n",
        "!pip install -q git+https://github.com/zphang/transformers@c3dc391 \n",
        "#Peft, is a method to load fine-tuned models -> https://huggingface.co/blog/peft\n",
        "!pip install -q git+https://github.com/huggingface/peft.git\n",
        "\n",
        "!pip -q install pynecone\n",
        "!npm -q install localtunnel\n",
        "clear_output()"
      ],
      "metadata": {
        "id": "X_pz8MuY84Qh"
      },
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title build & Initialize  - terminate instance for building errors\n",
        "# initialize and build \n",
        "!pc init\n",
        "# colab runs one cell at a time, running pc app should be a background process\n",
        "!pc run &>/content/logs.txt & \n",
        "\n",
        "# the following commands should be run in the background \n",
        "# accessing the output for these cells is important and should be coded\n",
        "!npx localtunnel --port 8000 &>/content/backend.txt &\n",
        "!npx localtunnel --port 3000 &>/content/frontend.txt &"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ClisncEH2ey4",
        "outputId": "9e67aa3a-f0f5-4901-f0d2-7599508074ab"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[?25l\u001b[2;36m[16:32:01]\u001b[0m\u001b[2;36m \u001b[0mInstalling bun\u001b[33m...\u001b[0m                                        \u001b]8;id=679231;file:///usr/local/lib/python3.9/dist-packages/pynecone/utils.py\u001b\\\u001b[2mutils.py\u001b[0m\u001b]8;;\u001b\\\u001b[2m:\u001b[0m\u001b]8;id=404654;file:///usr/local/lib/python3.9/dist-packages/pynecone/utils.py#440\u001b\\\u001b[2m440\u001b[0m\u001b]8;;\u001b\\\n",
            "######################################################################## 100.0%\n",
            "\u001b[2K\u001b[32m⠴\u001b[0m \u001b[1mInitializing content\u001b[0m\u001b[0;32mbun was installed successfully to \u001b[1;32m~/.bun/bin/bun \u001b[0m\n",
            "\n",
            "\u001b[2K\u001b[32m⠦\u001b[0m \u001b[1mInitializing content\u001b[0m\u001b[0;2mAdded \"~/.bun/bin\" to $PATH in \"~/.bashrc\" \u001b[0m\n",
            "\n",
            "\u001b[0;2mTo get started, run: \u001b[0m\n",
            "\n",
            "\u001b[1m source /root/.bashrc \u001b[0m\n",
            "\u001b[1m  bun --help \u001b[0m\n",
            "\u001b[2K\u001b[2;36m[16:32:03]\u001b[0m\u001b[2;36m \u001b[0mInitializing the web directory.                          \u001b]8;id=709969;file:///usr/local/lib/python3.9/dist-packages/pynecone/utils.py\u001b\\\u001b[2mutils.py\u001b[0m\u001b]8;;\u001b\\\u001b[2m:\u001b[0m\u001b]8;id=293690;file:///usr/local/lib/python3.9/dist-packages/pynecone/utils.py#421\u001b\\\u001b[2m421\u001b[0m\u001b]8;;\u001b\\\n",
            "\u001b[2K\u001b[2;36m          \u001b[0m\u001b[2;36m \u001b[0mInitializing the app directory.                          \u001b]8;id=931809;file:///usr/local/lib/python3.9/dist-packages/pynecone/utils.py\u001b\\\u001b[2mutils.py\u001b[0m\u001b]8;;\u001b\\\u001b[2m:\u001b[0m\u001b]8;id=719468;file:///usr/local/lib/python3.9/dist-packages/pynecone/utils.py#410\u001b\\\u001b[2m410\u001b[0m\u001b]8;;\u001b\\\n",
            "\u001b[2K\u001b[2;36m          \u001b[0m\u001b[2;36m \u001b[0m\u001b[1;32mFinished Initializing: content\u001b[0m                               \u001b]8;id=737367;file:///usr/local/lib/python3.9/dist-packages/pynecone/pc.py\u001b\\\u001b[2mpc.py\u001b[0m\u001b]8;;\u001b\\\u001b[2m:\u001b[0m\u001b]8;id=823912;file:///usr/local/lib/python3.9/dist-packages/pynecone/pc.py#53\u001b\\\u001b[2m53\u001b[0m\u001b]8;;\u001b\\\n",
            "\u001b[2K\u001b[32m⠸\u001b[0m \u001b[1mInitializing content\u001b[0m\n",
            "\u001b[1A\u001b[2K"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Warning \n",
        "you should have logs.txt opened to check how is pynecone performs in the background and also for debugging"
      ],
      "metadata": {
        "id": "12Px52BWcKdf"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Allocating ports \n",
        "\n",
        "#* wait for logs to be written, to fetch urls\n",
        "import time\n",
        "time.sleep(8)\n",
        "\n",
        "with open('frontend.txt') as f:\n",
        "  frontlines = f.readlines()  \n",
        "front_url = str(frontlines[1].split()[-1])\n",
        "# front_url"
      ],
      "metadata": {
        "id": "SHj64k-DF-Y-",
        "cellView": "form"
      },
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title  main app\n",
        "%%writefile /content/content/content.py\n",
        "\n",
        "import warnings\n",
        "warnings.filterwarnings(\"ignore\")\n",
        "\n",
        "####################################MODEL#####################################################\n",
        "\n",
        "def generate_prompt(instruction: str, input_ctxt: str = None) -> str:\n",
        "    if input_ctxt:\n",
        "        return f\"\"\"Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.\n",
        "\n",
        "### Instruction:\n",
        "{instruction}\n",
        "\n",
        "### Input:\n",
        "{input_ctxt}\n",
        "\n",
        "### Response:\"\"\"\n",
        "    else:\n",
        "        return f\"\"\"Below is an instruction that describes a task. Write a response that appropriately completes the request.\n",
        "\n",
        "### Instruction:\n",
        "{instruction}\n",
        "\n",
        "### Response:\"\"\"\n",
        "\n",
        "import torch\n",
        "from transformers import LLaMATokenizer, LLaMAForCausalLM, GenerationConfig\n",
        "\n",
        "tokenizer = LLaMATokenizer.from_pretrained(\"chainyo/alpaca-lora-7b\")\n",
        "model = LLaMAForCausalLM.from_pretrained(\n",
        "    \"chainyo/alpaca-lora-7b\",\n",
        "    load_in_8bit=True,\n",
        "    torch_dtype=torch.float16,\n",
        "    device_map=\"auto\",\n",
        ")\n",
        "\n",
        "model.eval()\n",
        "if torch.__version__ >= \"2\":\n",
        "    model = torch.compile(model)\n",
        "\n",
        "\n",
        "def response(instruction, generation_config, input_ctxt=None):\n",
        "  prompt = generate_prompt(instruction, input_ctxt)\n",
        "  input_ids = tokenizer(prompt, return_tensors=\"pt\").input_ids\n",
        "  input_ids = input_ids.to(model.device)\n",
        "\n",
        "  with torch.no_grad():\n",
        "      outputs = model.generate(\n",
        "          input_ids=input_ids,\n",
        "          generation_config=generation_config,\n",
        "          return_dict_in_generate=True,\n",
        "          output_scores=True,\n",
        "      )\n",
        "\n",
        "  answer = tokenizer.decode(outputs.sequences[0], skip_special_tokens=True)\n",
        "  return answer.split(\"### Response:\")[1].strip()\n",
        "\n",
        "\n",
        "####################################MODEL#####################################################\n",
        "\n",
        "\n",
        "import pynecone as pc\n",
        "\n",
        "\n",
        "class alpaca_chat(pc.State):\n",
        "    _question: str = \"\"\n",
        "    _answer: str = \"\"\n",
        "    input = \"\"\n",
        "\n",
        "    # ** check box for output\n",
        "    code: bool = False\n",
        "    code_lang: str = \"python\"\n",
        "    input_enable: str = False\n",
        "\n",
        "    # ** chech for answer\n",
        "    thinking = False\n",
        "\n",
        "    # * generation parameters\n",
        "    temp: float = 2.0\n",
        "    num_words: int = 45\n",
        "    #################### UI Changing params #############################\n",
        "    #TODO : connect them to the ui \n",
        "    \n",
        "    #################### Text Area ###############################\n",
        "    def set_question(self, question):\n",
        "        self._question = question\n",
        "\n",
        "    @pc.var\n",
        "    def get_question(self):\n",
        "        return self._question\n",
        "\n",
        "    ##################### button ################################\n",
        "\n",
        "    def process_answer(self):\n",
        "        self.thinking = True\n",
        "\n",
        "    def get_answer(self):\n",
        "        #TODO : Change your model \n",
        "        #TODO : handle inputs for instsructions \n",
        "        generation_config = GenerationConfig(\n",
        "          temperature = float(float(self.temp) / 10),\n",
        "          top_p=0.75,\n",
        "          top_k=40,\n",
        "          num_beams=4,\n",
        "          max_new_tokens= int(self.num_words))\n",
        "\n",
        "        self._answer = response(self._question, generation_config)\n",
        "        ## #####################\n",
        "        self.thinking = False\n",
        "\n",
        "    @pc.var\n",
        "    def view_answer(self):\n",
        "        return self._answer\n",
        "\n",
        "    ##################### generation parameters ##################\n",
        "\n",
        "    def code_output(self):\n",
        "        self.code = not (self.code)\n",
        "\n",
        "\n",
        "def index():\n",
        "    return pc.box(\n",
        "        pc.vstack(\n",
        "            # * Title\n",
        "            pc.center(\n",
        "                pc.heading(\n",
        "                    \"Alpaca-Q/A\",\n",
        "                    font_size=\"3.5em\",\n",
        "                    _hover={\"cursor\": \"pointer\"},\n",
        "                ),\n",
        "                padding_bottom=\"2em\",\n",
        "                width=\"75%\",\n",
        "                #padding_right=\"2em\",\n",
        "            ),\n",
        "            pc.hstack(\n",
        "                pc.vstack(\n",
        "                    # * Question spot\n",
        "                    pc.heading(\"instruction\", font_size=\"1.5em\", align_items=\"left\"),\n",
        "                    pc.text_area(\n",
        "                        placeholder=\"Ask Alpaca\",\n",
        "                        on_blur=alpaca_chat.set_question,\n",
        "                        width=\"100%\",\n",
        "                        background=\"white\",\n",
        "                    ),\n",
        "                    pc.divider(),\n",
        "                    pc.cond(\n",
        "                        alpaca_chat.input_enable,\n",
        "                        c1=pc.vstack(\n",
        "                            pc.heading(\"Input\", font_size=\"1.5em\"),\n",
        "                            pc.text_area(\n",
        "                                placeholder=\"input for Question\",\n",
        "                                on_blur=alpaca_chat.set_input,\n",
        "                                background=\"white\",\n",
        "                            ),\n",
        "                            width=\"100%\",\n",
        "                        ),\n",
        "                    ),\n",
        "                    pc.divider(),\n",
        "                    # * waiting for answer\n",
        "                    pc.cond(\n",
        "                        alpaca_chat.thinking,\n",
        "                        pc.vstack(\n",
        "                            pc.text(\"Alpaca is thinking \"),\n",
        "                            pc.progress(is_indeterminate=True, width=\"75%\"),\n",
        "                        ),\n",
        "                        # pc.text(\"Alapaca is done \"),\n",
        "                    ),\n",
        "                    # * Answer spot\n",
        "                    pc.cond(\n",
        "                        alpaca_chat.code,\n",
        "                        c1=pc.vstack(\n",
        "                            pc.text(\"you : \" + alpaca_chat.get_question,width=\"100%\",background=\"white\",),\n",
        "                            pc.text(\"Alpaca : \",width=\"100%\",background=\"white\",),\n",
        "                            pc.code_block(alpaca_chat.view_answer,width=\"100%\",language=alpaca_chat.code_lang,),\n",
        "                            width=\"100%\",background=\"white\",\n",
        "                        ),\n",
        "                        c2=pc.vstack(\n",
        "                            pc.text(\"you : \" + alpaca_chat.get_question, width=\"100%\", background=\"white\"),\n",
        "                            pc.text(\"Alpaca : \" + alpaca_chat.view_answer, width=\"100%\", background=\"white\"), width=\"100%\", background=\"white\"),\n",
        "                      ),\n",
        "                    pc.center(  # * chat button\n",
        "                        pc.button(\n",
        "                            \"Ask Alpaca\",\n",
        "                            width=\"100%\",\n",
        "                            on_click=[\n",
        "                                alpaca_chat.process_answer,\n",
        "                                alpaca_chat.get_answer,\n",
        "                            ],\n",
        "                            background=\"gray\",\n",
        "                        ),\n",
        "                    ),\n",
        "                    pc.accordion(\n",
        "                        pc.accordion_item(\n",
        "                            pc.accordion_button(\n",
        "                                pc.heading(\"Generation Settings\"),\n",
        "                                pc.accordion_icon(),\n",
        "                                reduce_motion=True,\n",
        "                            ),\n",
        "                            pc.accordion_panel(\n",
        "                                pc.center(\n",
        "                                    pc.vstack(\n",
        "                                        # * unique ability\n",
        "                                        pc.hstack(\n",
        "                                            pc.center(\n",
        "                                                pc.tooltip(\n",
        "                                                    pc.text(\n",
        "                                                        f\"Tempreture \", font_size=20\n",
        "                                                    ),\n",
        "                                                    label=\"This indicates unique-ability of genereation, set high for relatively close generation\",\n",
        "                                                ),\n",
        "                                                pc.spacer(),\n",
        "                                                pc.slider(\n",
        "                                                    default_value=2,\n",
        "                                                    on_change=alpaca_chat.set_temp,\n",
        "                                                    min_=1,\n",
        "                                                    max_=10,\n",
        "                                                    width=\"50%\",\n",
        "                                                ),\n",
        "                                                width=\"100%\",\n",
        "                                            ),\n",
        "                                            width=\"100%\",\n",
        "                                        ),\n",
        "                                        # * code output\n",
        "                                        pc.hstack(\n",
        "                                            pc.center(\n",
        "                                                pc.tooltip(\n",
        "                                                    pc.text(\n",
        "                                                        \"Code Output\", font_size=20\n",
        "                                                    ),\n",
        "                                                    label=\"enable to handle generated code correctly\",\n",
        "                                                ),\n",
        "                                                pc.spacer(),\n",
        "                                                pc.switch(\n",
        "                                                    on_change=alpaca_chat.set_code\n",
        "                                                ),\n",
        "                                                width=\"100%\",\n",
        "                                            ),\n",
        "                                            width=\"100%\",\n",
        "                                        ),\n",
        "                                        # * Input enable for instruction like summarize:\n",
        "                                        pc.hstack(\n",
        "                                            pc.center(\n",
        "                                                pc.tooltip(\n",
        "                                                    pc.text(\n",
        "                                                        \"Instruction's input - Not Working yet\",\n",
        "                                                        font_size=20,\n",
        "                                                    ),\n",
        "                                                    label=\"Enable when your task includes inputs, ex : Summarize \",\n",
        "                                                ),\n",
        "                                                pc.spacer(),\n",
        "                                                pc.switch(\n",
        "                                                    on_change=alpaca_chat.set_input_enable\n",
        "                                                ),\n",
        "                                                width=\"100%\",\n",
        "                                            ),\n",
        "                                            width=\"100%\",\n",
        "                                        ),\n",
        "                                        # * number of generated words\n",
        "                                        pc.hstack(\n",
        "                                            pc.center(\n",
        "                                                pc.tooltip(\n",
        "                                                    pc.text(\n",
        "                                                        \"Max number of tokens\",\n",
        "                                                        font_size=20,\n",
        "                                                    ),\n",
        "                                                    label=\"Max Number of generated words\",\n",
        "                                                ),\n",
        "                                                pc.spacer(),\n",
        "                                                pc.number_input(\n",
        "                                                    defaultValue=64,\n",
        "                                                    on_change=alpaca_chat.set_num_words,\n",
        "                                                    width=\"50%\",\n",
        "                                                    background=\"white\",\n",
        "                                                    max_ = 250,\n",
        "                                                    min_ = 10,\n",
        "                                                ),\n",
        "                                                width=\"100%\",\n",
        "                                            ),\n",
        "                                            width=\"100%\",\n",
        "                                        ),\n",
        "                                        # * language output horizontal stack\n",
        "                                        pc.hstack(\n",
        "                                            pc.center(\n",
        "                                                pc.tooltip(\n",
        "                                                    pc.text(\n",
        "                                                        \"code language\", font_size=20\n",
        "                                                    ),\n",
        "                                                    label=\"Write the type of language\",\n",
        "                                                ),\n",
        "                                                pc.spacer(),\n",
        "                                                pc.input(\n",
        "                                                    defaultValue=\"python\",\n",
        "                                                    on_blur=alpaca_chat.set_code_lang,\n",
        "                                                    width=\"50%\",\n",
        "                                                    background=\"white\",\n",
        "                                                ),\n",
        "                                                # pc.switch(on_change=alpaca_chat.set_code),\n",
        "                                                width=\"100%\",\n",
        "                                            ),\n",
        "                                            width=\"100%\",\n",
        "                                        ),\n",
        "                                        pc.alert(\n",
        "                                            pc.alert_icon(),\n",
        "                                            pc.alert_title(\n",
        "                                                \"Refresh if it took longer than intended, don't worry, your answer'll be there\"\n",
        "                                            ),\n",
        "                                            status=\"warning\",\n",
        "                                        ),\n",
        "                                    ),\n",
        "                                    width=\"100%\",\n",
        "                                ),\n",
        "                            ),\n",
        "                        ),\n",
        "                        width=\"100%\",\n",
        "                    ),\n",
        "                    width=\"100%\",\n",
        "                    padding_bottom=\"4em\",\n",
        "                ),\n",
        "                width=\"80%\",\n",
        "                align_items=\"bottom\",\n",
        "            ),\n",
        "            width=\"100%\",\n",
        "            height=\"100%\",\n",
        "            padding_top=\"2em\",\n",
        "            # background=\"radial-gradient(circle at 22% 11%,rgba(62, 180, 137,.20),hsla(0,0%,100%,0) 19%),radial-gradient(circle at 82% 25%,rgba(33,150,243,.18),hsla(0,0%,100%,0) 35%),radial-gradient(circle at 25% 61%,rgba(250, 128, 114, .28),hsla(0,0%,100%,0) 55%)\",\n",
        "            # background = 'white',\n",
        "        ),\n",
        "        width=\"100%\",\n",
        "        height=\"100%\",\n",
        "        padding_top=\"2em\",\n",
        "        background=\"radial-gradient(circle at 22% 11%,rgba(62, 180, 137,.20),hsla(0,0%,100%,0) 19%),radial-gradient(circle at 82% 25%,rgba(33,150,243,.18),hsla(0,0%,100%,0) 35%),radial-gradient(circle at 25% 61%,rgba(250, 128, 114, .28),hsla(0,0%,100%,0) 55%)\",\n",
        "    )\n",
        "\n",
        "\n",
        "app = pc.App(state=alpaca_chat)\n",
        "app.add_page(index)\n",
        "app.compile()\n"
      ],
      "metadata": {
        "id": "P3Ucf6P1FSOk",
        "outputId": "55a29cc5-4966-4b17-8f6c-0c9afdd0e9a5",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Overwriting /content/content/content.py\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#@title recorrect configuations\n",
        "%%writefile /content/pcconfig.py\n",
        "\n",
        "import pynecone as pc\n",
        "\n",
        "with open('backend.txt') as f:\n",
        "  lines = f.readlines()\n",
        "back_url = str(lines[1].split()[-1])\n",
        "\n",
        "config = pc.Config(\n",
        "    app_name=\"content\",\n",
        "    api_url=back_url, # connecting the backend\n",
        "    bun_path=\"/content/.bun/bin/bun\", # recorrect bun path\n",
        "    db_url=\"sqlite:///pynecone.db\",\n",
        "    telemetry_enabled=False,\n",
        "    port = 8000\n",
        ")"
      ],
      "metadata": {
        "id": "QSRFAUbiFjI9",
        "outputId": "65a286ec-f706-4d2e-ebeb-bc2e263c0842",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "cellView": "form"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Overwriting /content/pcconfig.py\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Don't forget to rerun"
      ],
      "metadata": {
        "id": "JrNSh6GqATaH"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "print(\"click here to view app\", front_url)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kETxMjKMHY65",
        "outputId": "048fe34c-8ad8-4f50-9a96-6a22dd56cb3e"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "click here to view app https://afraid-masks-sip-35-247-82-41.loca.lt\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Just CLick to Continue"
      ],
      "metadata": {
        "id": "MdByU7ElQb4o"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#@title app-cell, might take few seconds\n",
        "import ipywidgets as widgets\n",
        "from IPython.display import IFrame\n",
        "IFrame(src=front_url, width='100%', height='600px')"
      ],
      "metadata": {
        "id": "b6seAtaZZShM",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 621
        },
        "outputId": "f288ca4c-0900-491a-9cc0-00f1ed248dff"
      },
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<IPython.lib.display.IFrame at 0x7f1850688850>"
            ],
            "text/html": [
              "\n",
              "        <iframe\n",
              "            width=\"100%\"\n",
              "            height=\"600px\"\n",
              "            src=\"https://afraid-masks-sip-35-247-82-41.loca.lt\"\n",
              "            frameborder=\"0\"\n",
              "            allowfullscreen\n",
              "        ></iframe>\n",
              "        "
            ]
          },
          "metadata": {},
          "execution_count": 16
        }
      ]
    }
  ]
}