{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "PAM8swqfl3YC"
      },
      "outputs": [],
      "source": [
        "import itertools\n",
        "import lzma\n",
        "import regex as re\n",
        "import torch\n",
        "from torch import nn\n",
        "from torch.utils.data import IterableDataset, DataLoader\n",
        "from torchtext.vocab import build_vocab_from_iterator\n",
        "import pickle\n",
        "import os"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "52BQle50l92y",
        "outputId": "1f98398d-f385-4711-c2b7-3abe7418fbdb"
      },
      "outputs": [],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "id": "PNb3_zqUl3YD"
      },
      "source": [
        "### Definitions"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "b_6d7n2al3YE"
      },
      "outputs": [],
      "source": [
        "def clean_line(line: str):\n",
        "    separated = line.split('\\t')\n",
        "    prefix = separated[6].replace(r'\\n', ' ').strip()\n",
        "    suffix = separated[7].replace(r'\\n', ' ').strip()\n",
        "    return prefix + ' ' + suffix\n",
        "\n",
        "def get_words_from_line(line):\n",
        "    line = clean_line(line)\n",
        "    for word in line.split():\n",
        "        yield word\n",
        "\n",
        "def get_word_lines_from_file(file_name):\n",
        "    with lzma.open(file_name, mode='rt', encoding='utf-8') as fid:\n",
        "        for line in fid:\n",
        "            yield get_words_from_line(line)\n",
        "\n",
        "\n",
        "def look_ahead_iterator(gen):\n",
        "    prev = None\n",
        "    for item in gen:\n",
        "        if prev is not None:\n",
        "            yield (prev, item)\n",
        "        prev = item\n",
        "\n",
        "def predict(word: str, num_of_top: str) -> str:\n",
        "    ixs = torch.tensor(vocab.forward([word])).to(device)\n",
        "    out = model(ixs)\n",
        "    top = torch.topk(out[0], num_of_top)\n",
        "    top_indices = top.indices.tolist()\n",
        "    top_probs = top.values.tolist()\n",
        "    top_words = vocab.lookup_tokens(top_indices)\n",
        "    zipped = list(zip(top_words, top_probs))\n",
        "    if '<unk>' in [element[0] for element in zipped]:\n",
        "        zipped = [(element[0] if element[0] != '<unk>' else '', element[1]) for element in zipped]\n",
        "        zipped[-1] = ('', zipped[-1][1])\n",
        "    else:\n",
        "        zipped[-1] = ('', zipped[-1][1])\n",
        "    return ' '.join([f'{element[0]}:{element[1]}' for element in zipped])\n",
        "\n",
        "def execute(path):\n",
        "    with lzma.open(f'{path}/in.tsv.xz', 'rt', encoding='utf-8') as f, \\\n",
        "        open(f'{path}/out.tsv', 'w', encoding='utf-8') as out:\n",
        "        for line in f:\n",
        "            prefix = line.split('\\t')[6]\n",
        "            left = prefix.replace(r'\\n', ' ').split()[-1]\n",
        "            result = predict(left, num_of_top)\n",
        "            out.write(f\"{result}\\n\")\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ZfV8fDhyl3YF"
      },
      "outputs": [],
      "source": [
        "class Bigrams(IterableDataset):\n",
        "  def __init__(self, text_file, vocabulary_size):\n",
        "      self.vocab = vocab\n",
        "      self.vocab.set_default_index(self.vocab['<unk>'])\n",
        "      self.vocabulary_size = vocabulary_size\n",
        "      self.text_file = text_file\n",
        "\n",
        "  def __iter__(self):\n",
        "     return look_ahead_iterator(\n",
        "         (self.vocab[t] for t in itertools.chain.from_iterable(get_word_lines_from_file(self.text_file))))\n",
        "\n",
        "        \n",
        "class SimpleBigramNeuralLanguageModel(nn.Module):\n",
        "  def __init__(self, vocabulary_size, embedding_size):\n",
        "      super(SimpleBigramNeuralLanguageModel, self).__init__()\n",
        "      self.model = nn.Sequential(\n",
        "          nn.Embedding(vocabulary_size, embedding_size),\n",
        "          nn.Linear(embedding_size, vocabulary_size),\n",
        "          nn.Softmax()\n",
        "      )\n",
        "\n",
        "  def forward(self, x):\n",
        "      return self.model(x)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "W0O6U62El3YG"
      },
      "source": [
        "### Parameters"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "eUS-U3_6l3YG"
      },
      "outputs": [],
      "source": [
        "vocab_size = 10000\n",
        "embed_size = 250\n",
        "batch_size = 5000\n",
        "num_of_top = 10"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CPeVRcYZl3YG"
      },
      "source": [
        "### Vocabulary building"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "4wBx0OTal3YH"
      },
      "outputs": [],
      "source": [
        "if os.path.exists('./vocabulary.pickle'):\n",
        "    with open('vocabulary.pickle', 'rb') as handle:\n",
        "        vocab = pickle.load(handle)\n",
        "else:\n",
        "    vocab = build_vocab_from_iterator(\n",
        "    get_word_lines_from_file('./drive/MyDrive/ColabNotebooks/america/train/in.tsv.xz'),\n",
        "    max_tokens = vocab_size,\n",
        "    specials = ['<unk>'])\n",
        "\n",
        "    with open('vocabulary.pickle', 'wb') as handle:\n",
        "        pickle.dump(vocab, handle, protocol=pickle.HIGHEST_PROTOCOL)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oVMVipnhl3YH",
        "outputId": "e588d083-be33-4dce-e61c-b26e664b2c5f"
      },
      "outputs": [],
      "source": [
        "vocab.lookup_tokens([0, 1, 2, 3, 4, 4500])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1gJscHJUl3YI"
      },
      "source": [
        "### Training"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "W_OjBUInl3YI"
      },
      "outputs": [],
      "source": [
        "model = SimpleBigramNeuralLanguageModel(vocab_size, embed_size)\n",
        "vocab.set_default_index(vocab['<unk>'])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "f373Kduzl3YI",
        "outputId": "0e6100df-d00e-4be3-83c4-cd2f671041e6"
      },
      "outputs": [],
      "source": [
        "#uczenie\n",
        "from torch.utils.data import DataLoader\n",
        "\n",
        "device = 'cuda'\n",
        "train_dataset = Bigrams('./drive/MyDrive/ColabNotebooks/america/train/in.tsv.xz', vocab_size)\n",
        "model = SimpleBigramNeuralLanguageModel(vocab_size, embed_size).to(device)\n",
        "data = DataLoader(train_dataset, batch_size=batch_size)\n",
        "optimizer = torch.optim.Adam(model.parameters())\n",
        "\n",
        "#funkcja kosztu\n",
        "criterion = torch.nn.NLLLoss()\n",
        "\n",
        "model.train()\n",
        "step = 0\n",
        "for x, y in data:\n",
        "   x = x.to(device)\n",
        "   y = y.to(device)\n",
        "   optimizer.zero_grad()\n",
        "   ypredicted = model(x)\n",
        "   loss = criterion(torch.log(ypredicted), y)\n",
        "   if step % 100 == 0:\n",
        "      print(step, loss)\n",
        "   step += 1\n",
        "   loss.backward()\n",
        "   optimizer.step()\n",
        "\n",
        "torch.save(model.state_dict(), 'model1.bin')\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IHrqskyXl3YI"
      },
      "source": [
        "### Evaluation"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yK3oK65fl3YI",
        "outputId": "ee3e64a6-c361-4e96-cad0-82f2950827ca"
      },
      "outputs": [],
      "source": [
        "model = SimpleBigramNeuralLanguageModel(vocab_size, embed_size).to(device)\n",
        "model.load_state_dict(torch.load('model1.bin'))\n",
        "model.eval()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ihz4Px0bl3YJ"
      },
      "outputs": [],
      "source": [
        "execute('./drive/MyDrive/ColabNotebooks/america/dev-0')\n",
        "execute('./drive/MyDrive/ColabNotebooks/america/test-A')"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "provenance": []
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.10.8"
    },
    "orig_nbformat": 4
  },
  "nbformat": 4,
  "nbformat_minor": 0
}