Text_Prediction_Using_LSTM.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "gpuType": "T4"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "# Importing Dependencies "
      ],
      "metadata": {
        "id": "4SFxa8PgAF9u"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "wPi7RbukPh-g"
      },
      "outputs": [],
      "source": [
        "\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense, Dropout, LSTM\n",
        "from keras.optimizers import RMSprop\n",
        "import numpy as np\n",
        "import random\n",
        "import sys\n"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#Loading the text\n",
        "filename = \"/content/chankya neeti.txt\"\n",
        "raw_text = open(filename, 'r', encoding='utf-8').read()\n",
        "raw_text = raw_text.lower()\n",
        "print(raw_text[0:1000])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LF6mIxZ7ALLD",
        "outputId": "7aae96e0-e2c2-4618-fe29-e8bd44fca6fd"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "“1.1 सर्वशक्तिमान तीनो लोको के स्वामी श्री विष्णु भगवान को शीश नवाकर मै अनेक शास्त्रों से निकाले गए राजनीति सार के तत्व को जन कल्याण हेतु समाज के सम्मुख रखता हूं”\n",
            "“1.2 इस राजनीति शास्त्र का विधिपूर्वक अध्ययन करके यह जाना जा सकता है कि कौनसा कार्य करना चाहिए और कौनसा कार्य नहीं करना चाहिए। यह जानकर वह एक प्रकार से धर्मोपदेश प्राप्त करता है कि किस कार्य के करने से अच्छा परिणाम निकलेगा और किससे बुरा। उसे अच्छे बुरे का ज्ञान हो जाता है”\n",
            "“1.3 इसलिए जनता की भलाई के लिए, मैं वह बात कहूंगा, जो जब समझ में आएगी, तब उनके उचित परिप्रेक्ष्य में चीजों की समझ पैदा होगी”\n",
            "“1.4 यहां तक ​​कि एक पंडित एक मूर्ख शिष्य को निर्देश देता है, एक दुष्ट पत्नी को बनाए रखने से, और दुखी के साथ अत्यधिक परिचित होने से”\n",
            "“1.5 दुष्ट स्त्री, छल करने वाला मित्र, पलटकर कर तीखा जवाब देने वाला नौकर तथा जिस घर में सांप रहता हो, उस घर में निवास करने वाले गृहस्वामी की मौत में संशय न करे। वह निश्चित मृत्यु को प्राप्त होता है। – चाणक्य नीति”\n",
            "“1.6 विपत्ति के समय काम आने वाले धन की रक्षा करे। धन से स्त्री की रक्षा करे और अपनी रक्षा ध\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Pre-Processing"
      ],
      "metadata": {
        "id": "kRnqFLIjC3_T"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#CLEAN TEXT\n",
        "#Remove numbers\n",
        "raw_text = ''.join(c for c in raw_text if not c.isdigit())\n",
        "\n",
        "#How many total characters do we have in our training text?\n",
        "chars = sorted(list(set(raw_text))) #List of every character\n",
        "\n",
        "#Character sequences must be encoded as integers. \n",
        "#Each unique character will be assigned an integer value. \n",
        "#Create a dictionary of characters mapped to integer values\n",
        "char_to_int = dict((c, i) for i, c in enumerate(chars))\n",
        "\n",
        "#Do the reverse so we can print our predictions in characters and not integers\n",
        "int_to_char = dict((i, c) for i, c in enumerate(chars))"
      ],
      "metadata": {
        "id": "xBwC_thwBiMe"
      },
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Data Exploration"
      ],
      "metadata": {
        "id": "_mU-NqZrHG2U"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# summarize the data\n",
        "n_chars = len(raw_text)\n",
        "n_vocab = len(chars)\n",
        "print(\"Total Characters in the text; corpus length: \", n_chars)\n",
        "print(\"Total Vocab: \", n_vocab)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EbLijafDCwcN",
        "outputId": "76a1a878-da33-42e6-f559-1a1b557c59a1"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Total Characters in the text; corpus length:  45245\n",
            "Total Vocab:  79\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from re import X\n",
        "#Now that we have characters we can create input/output sequences for training\n",
        "#Remember that for LSTM input and output can be sequences... hence the term seq2seq\n",
        "\n",
        "\n",
        "seq_length = 60  #Length of each input sequence\n",
        "step = 10   #Instead of moving 1 letter at a time, here we are trying to skipping a few. \n",
        "sentences = []    # X values (Sentences)\n",
        "next_chars = []   # Y values. The character that follows the sentence defined as X\n",
        "for i in range(0, n_chars - seq_length, step):  #step=1 means each sentence is offset just by a single letter\n",
        "    sentences.append(raw_text[i: i + seq_length])  #Sequence in\n",
        "    next_chars.append(raw_text[i + seq_length])  #Sequence out\n",
        "n_patterns = len(sentences)    \n",
        "print('Number of sequences:', n_patterns)\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "azSryvMEHJe0",
        "outputId": "5cb3be4d-474f-49d5-d105-2b201e3ec088"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Number of sequences: 4519\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Just like time series, X is the sequence / sentence and y is the next value that comes after the sentence... \n",
        "reshape input to be [samples, time steps, features]\n",
        "time steps = sequence length, features = numbers of characters in our vocab (n_vocab)\n",
        "Vectorize all sentences: there are n_patterns sentences.\n",
        "For each sentence we have n_vocab characters available for seq_length\n",
        "Vectorization returns a vector for all sentences indicating the presence or absence of a character. "
      ],
      "metadata": {
        "id": "ZjIc852PJkFD"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "x = np.zeros((len(sentences), seq_length, n_vocab), dtype=np.bool)\n",
        "y = np.zeros((len(sentences), n_vocab), dtype=np.bool)\n",
        "for i, sentence in enumerate(sentences):\n",
        "    for t, char in enumerate(sentence):\n",
        "        x[i, t, char_to_int[char]] = 1\n",
        "    y[i, char_to_int[next_chars[i]]] = 1\n",
        "    \n",
        "print(x.shape)\n",
        "print(y.shape)\n",
        "\n",
        "#print(y[0:10]) commenting this because it will take too much space it is like one hot encoding for first 10 charecters"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jXqJb_ZxMFdj",
        "outputId": "f514fc3a-985a-412d-c18e-39e2406cb843"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "<ipython-input-6-07e6c2a1329d>:1: DeprecationWarning: `np.bool` is a deprecated alias for the builtin `bool`. To silence this warning, use `bool` by itself. Doing this will not modify any behavior and is safe. If you specifically wanted the numpy scalar type, use `np.bool_` here.\n",
            "Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations\n",
            "  x = np.zeros((len(sentences), seq_length, n_vocab), dtype=np.bool)\n",
            "<ipython-input-6-07e6c2a1329d>:2: DeprecationWarning: `np.bool` is a deprecated alias for the builtin `bool`. To silence this warning, use `bool` by itself. Doing this will not modify any behavior and is safe. If you specifically wanted the numpy scalar type, use `np.bool_` here.\n",
            "Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations\n",
            "  y = np.zeros((len(sentences), n_vocab), dtype=np.bool)\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(4519, 60, 79)\n",
            "(4519, 79)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Single LSTM"
      ],
      "metadata": {
        "id": "cZrnl6ceRjst"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#Basic model with one LSTM\n",
        "# build the model: a single LSTM\n",
        "\n",
        "model = Sequential()\n",
        "model.add(LSTM(128, input_shape=(seq_length, n_vocab)))\n",
        "model.add(Dense(n_vocab, activation='softmax'))\n",
        "\n",
        "optimizer = RMSprop(lr=0.01)\n",
        "model.compile(loss='categorical_crossentropy', optimizer=optimizer)\n",
        "model.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tju0ZZrpRJF_",
        "outputId": "05a90b26-4a18-47e1-d2fd-e4b13e40ed21"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " lstm (LSTM)                 (None, 128)               106496    \n",
            "                                                                 \n",
            " dense (Dense)               (None, 79)                10191     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 116,687\n",
            "Trainable params: 116,687\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/keras/optimizers/legacy/rmsprop.py:143: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead.\n",
            "  super().__init__(name, **kwargs)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# define the checkpoint\n",
        "from keras.callbacks import ModelCheckpoint\n",
        "\n",
        "filepath=\"saved_weights/saved_weights-{epoch:02d}-{loss:.4f}.hdf5\"\n",
        "checkpoint = ModelCheckpoint(filepath, monitor='loss', verbose=1, save_best_only=True, mode='min')\n",
        "\n",
        "callbacks_list = [checkpoint]\n",
        "\n",
        "\n",
        "# Fit the model\n",
        "\n",
        "history = model.fit(x, y,\n",
        "          batch_size=128,\n",
        "          epochs=60,   \n",
        "          callbacks=callbacks_list)\n",
        "\n",
        "model.save('my_saved_weights_hindi_60epochs.h5')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "F7tLUEdMRZy2",
        "outputId": "5711bc45-b5cf-402e-d2ee-c62dca2c8e02"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 3.4910\n",
            "Epoch 1: loss improved from inf to 3.46150, saving model to saved_weights/saved_weights-01-3.4615.hdf5\n",
            "36/36 [==============================] - 7s 11ms/step - loss: 3.4615\n",
            "Epoch 2/60\n",
            "29/36 [=======================>......] - ETA: 0s - loss: 2.8756\n",
            "Epoch 2: loss improved from 3.46150 to 2.85099, saving model to saved_weights/saved_weights-02-2.8510.hdf5\n",
            "36/36 [==============================] - 0s 8ms/step - loss: 2.8510\n",
            "Epoch 3/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 2.5203\n",
            "Epoch 3: loss improved from 2.85099 to 2.50942, saving model to saved_weights/saved_weights-03-2.5094.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 2.5094\n",
            "Epoch 4/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 2.3287\n",
            "Epoch 4: loss improved from 2.50942 to 2.32549, saving model to saved_weights/saved_weights-04-2.3255.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 2.3255\n",
            "Epoch 5/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 2.1732\n",
            "Epoch 5: loss improved from 2.32549 to 2.17878, saving model to saved_weights/saved_weights-05-2.1788.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 2.1788\n",
            "Epoch 6/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 2.0296\n",
            "Epoch 6: loss improved from 2.17878 to 2.03618, saving model to saved_weights/saved_weights-06-2.0362.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 2.0362\n",
            "Epoch 7/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 1.8882\n",
            "Epoch 7: loss improved from 2.03618 to 1.89348, saving model to saved_weights/saved_weights-07-1.8935.hdf5\n",
            "36/36 [==============================] - 0s 8ms/step - loss: 1.8935\n",
            "Epoch 8/60\n",
            "31/36 [========================>.....] - ETA: 0s - loss: 1.7695\n",
            "Epoch 8: loss improved from 1.89348 to 1.75593, saving model to saved_weights/saved_weights-08-1.7559.hdf5\n",
            "36/36 [==============================] - 0s 10ms/step - loss: 1.7559\n",
            "Epoch 9/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 1.6021\n",
            "Epoch 9: loss improved from 1.75593 to 1.61142, saving model to saved_weights/saved_weights-09-1.6114.hdf5\n",
            "36/36 [==============================] - 0s 9ms/step - loss: 1.6114\n",
            "Epoch 10/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 1.4300\n",
            "Epoch 10: loss improved from 1.61142 to 1.43124, saving model to saved_weights/saved_weights-10-1.4312.hdf5\n",
            "36/36 [==============================] - 0s 9ms/step - loss: 1.4312\n",
            "Epoch 11/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 1.2614\n",
            "Epoch 11: loss improved from 1.43124 to 1.26397, saving model to saved_weights/saved_weights-11-1.2640.hdf5\n",
            "36/36 [==============================] - 0s 9ms/step - loss: 1.2640\n",
            "Epoch 12/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 1.0981\n",
            "Epoch 12: loss improved from 1.26397 to 1.09807, saving model to saved_weights/saved_weights-12-1.0981.hdf5\n",
            "36/36 [==============================] - 0s 8ms/step - loss: 1.0981\n",
            "Epoch 13/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.9328\n",
            "Epoch 13: loss improved from 1.09807 to 0.93285, saving model to saved_weights/saved_weights-13-0.9328.hdf5\n",
            "36/36 [==============================] - 0s 9ms/step - loss: 0.9328\n",
            "Epoch 14/60\n",
            "30/36 [========================>.....] - ETA: 0s - loss: 0.7725\n",
            "Epoch 14: loss improved from 0.93285 to 0.77572, saving model to saved_weights/saved_weights-14-0.7757.hdf5\n",
            "36/36 [==============================] - 0s 10ms/step - loss: 0.7757\n",
            "Epoch 15/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.6561\n",
            "Epoch 15: loss improved from 0.77572 to 0.65423, saving model to saved_weights/saved_weights-15-0.6542.hdf5\n",
            "36/36 [==============================] - 0s 9ms/step - loss: 0.6542\n",
            "Epoch 16/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.5337\n",
            "Epoch 16: loss improved from 0.65423 to 0.53742, saving model to saved_weights/saved_weights-16-0.5374.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.5374\n",
            "Epoch 17/60\n",
            "31/36 [========================>.....] - ETA: 0s - loss: 0.4543\n",
            "Epoch 17: loss improved from 0.53742 to 0.46148, saving model to saved_weights/saved_weights-17-0.4615.hdf5\n",
            "36/36 [==============================] - 0s 8ms/step - loss: 0.4615\n",
            "Epoch 18/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.3833\n",
            "Epoch 18: loss improved from 0.46148 to 0.38505, saving model to saved_weights/saved_weights-18-0.3851.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.3851\n",
            "Epoch 19/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 0.3375\n",
            "Epoch 19: loss improved from 0.38505 to 0.33973, saving model to saved_weights/saved_weights-19-0.3397.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.3397\n",
            "Epoch 20/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.2905\n",
            "Epoch 20: loss improved from 0.33973 to 0.29585, saving model to saved_weights/saved_weights-20-0.2959.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.2959\n",
            "Epoch 21/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.2422\n",
            "Epoch 21: loss improved from 0.29585 to 0.25343, saving model to saved_weights/saved_weights-21-0.2534.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.2534\n",
            "Epoch 22/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.2320\n",
            "Epoch 22: loss improved from 0.25343 to 0.23620, saving model to saved_weights/saved_weights-22-0.2362.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.2362\n",
            "Epoch 23/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.2129\n",
            "Epoch 23: loss improved from 0.23620 to 0.21506, saving model to saved_weights/saved_weights-23-0.2151.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.2151\n",
            "Epoch 24/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.1803\n",
            "Epoch 24: loss improved from 0.21506 to 0.18575, saving model to saved_weights/saved_weights-24-0.1857.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1857\n",
            "Epoch 25/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.1710\n",
            "Epoch 25: loss improved from 0.18575 to 0.17839, saving model to saved_weights/saved_weights-25-0.1784.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1784\n",
            "Epoch 26/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 0.1778\n",
            "Epoch 26: loss did not improve from 0.17839\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1792\n",
            "Epoch 27/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.1588\n",
            "Epoch 27: loss improved from 0.17839 to 0.16087, saving model to saved_weights/saved_weights-27-0.1609.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1609\n",
            "Epoch 28/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.1423\n",
            "Epoch 28: loss improved from 0.16087 to 0.14738, saving model to saved_weights/saved_weights-28-0.1474.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1474\n",
            "Epoch 29/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.1484\n",
            "Epoch 29: loss did not improve from 0.14738\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1532\n",
            "Epoch 30/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.1390\n",
            "Epoch 30: loss improved from 0.14738 to 0.14106, saving model to saved_weights/saved_weights-30-0.1411.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1411\n",
            "Epoch 31/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.1590\n",
            "Epoch 31: loss did not improve from 0.14106\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1643\n",
            "Epoch 32/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.1409\n",
            "Epoch 32: loss did not improve from 0.14106\n",
            "36/36 [==============================] - 0s 8ms/step - loss: 0.1416\n",
            "Epoch 33/60\n",
            "31/36 [========================>.....] - ETA: 0s - loss: 0.1276\n",
            "Epoch 33: loss improved from 0.14106 to 0.13341, saving model to saved_weights/saved_weights-33-0.1334.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1334\n",
            "Epoch 34/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.1253\n",
            "Epoch 34: loss improved from 0.13341 to 0.12533, saving model to saved_weights/saved_weights-34-0.1253.hdf5\n",
            "36/36 [==============================] - 0s 8ms/step - loss: 0.1253\n",
            "Epoch 35/60\n",
            "31/36 [========================>.....] - ETA: 0s - loss: 0.1195\n",
            "Epoch 35: loss improved from 0.12533 to 0.12488, saving model to saved_weights/saved_weights-35-0.1249.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1249\n",
            "Epoch 36/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.1334\n",
            "Epoch 36: loss did not improve from 0.12488\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1385\n",
            "Epoch 37/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.1142\n",
            "Epoch 37: loss improved from 0.12488 to 0.11550, saving model to saved_weights/saved_weights-37-0.1155.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1155\n",
            "Epoch 38/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.1314\n",
            "Epoch 38: loss did not improve from 0.11550\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1308\n",
            "Epoch 39/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.0995\n",
            "Epoch 39: loss improved from 0.11550 to 0.11095, saving model to saved_weights/saved_weights-39-0.1110.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1110\n",
            "Epoch 40/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.1190\n",
            "Epoch 40: loss did not improve from 0.11095\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1203\n",
            "Epoch 41/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.1229\n",
            "Epoch 41: loss did not improve from 0.11095\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1224\n",
            "Epoch 42/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.1001\n",
            "Epoch 42: loss improved from 0.11095 to 0.10386, saving model to saved_weights/saved_weights-42-0.1039.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1039\n",
            "Epoch 43/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.1113\n",
            "Epoch 43: loss did not improve from 0.10386\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1182\n",
            "Epoch 44/60\n",
            "31/36 [========================>.....] - ETA: 0s - loss: 0.1019\n",
            "Epoch 44: loss did not improve from 0.10386\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1069\n",
            "Epoch 45/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.0958\n",
            "Epoch 45: loss improved from 0.10386 to 0.09963, saving model to saved_weights/saved_weights-45-0.0996.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.0996\n",
            "Epoch 46/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.1154\n",
            "Epoch 46: loss did not improve from 0.09963\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1163\n",
            "Epoch 47/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.0924\n",
            "Epoch 47: loss improved from 0.09963 to 0.09504, saving model to saved_weights/saved_weights-47-0.0950.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.0950\n",
            "Epoch 48/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.0909\n",
            "Epoch 48: loss improved from 0.09504 to 0.09198, saving model to saved_weights/saved_weights-48-0.0920.hdf5\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.0920\n",
            "Epoch 49/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.0938\n",
            "Epoch 49: loss did not improve from 0.09198\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.0965\n",
            "Epoch 50/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.0963\n",
            "Epoch 50: loss did not improve from 0.09198\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.0971\n",
            "Epoch 51/60\n",
            "31/36 [========================>.....] - ETA: 0s - loss: 0.0998\n",
            "Epoch 51: loss did not improve from 0.09198\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1052\n",
            "Epoch 52/60\n",
            "30/36 [========================>.....] - ETA: 0s - loss: 0.1018\n",
            "Epoch 52: loss did not improve from 0.09198\n",
            "36/36 [==============================] - 0s 7ms/step - loss: 0.1101\n",
            "Epoch 53/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 0.1034\n",
            "Epoch 53: loss did not improve from 0.09198\n",
            "36/36 [==============================] - 0s 8ms/step - loss: 0.1052\n",
            "Epoch 54/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 0.0812\n",
            "Epoch 54: loss improved from 0.09198 to 0.08336, saving model to saved_weights/saved_weights-54-0.0834.hdf5\n",
            "36/36 [==============================] - 0s 9ms/step - loss: 0.0834\n",
            "Epoch 55/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 0.0834\n",
            "Epoch 55: loss improved from 0.08336 to 0.08202, saving model to saved_weights/saved_weights-55-0.0820.hdf5\n",
            "36/36 [==============================] - 0s 9ms/step - loss: 0.0820\n",
            "Epoch 56/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.0877\n",
            "Epoch 56: loss did not improve from 0.08202\n",
            "36/36 [==============================] - 0s 9ms/step - loss: 0.0873\n",
            "Epoch 57/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.0724\n",
            "Epoch 57: loss improved from 0.08202 to 0.07196, saving model to saved_weights/saved_weights-57-0.0720.hdf5\n",
            "36/36 [==============================] - 0s 9ms/step - loss: 0.0720\n",
            "Epoch 58/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.0760\n",
            "Epoch 58: loss did not improve from 0.07196\n",
            "36/36 [==============================] - 0s 9ms/step - loss: 0.0766\n",
            "Epoch 59/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 0.0885\n",
            "Epoch 59: loss did not improve from 0.07196\n",
            "36/36 [==============================] - 0s 9ms/step - loss: 0.0925\n",
            "Epoch 60/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 0.0890\n",
            "Epoch 60: loss did not improve from 0.07196\n",
            "36/36 [==============================] - 0s 9ms/step - loss: 0.0917\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#Tracking loos function \n",
        "from matplotlib import pyplot as plt\n",
        "#plot the training and validation accuracy and loss at each epoch\n",
        "loss = history.history['loss']\n",
        "epochs = range(1, len(loss) + 1)\n",
        "plt.plot(epochs, loss, 'y', label='Training loss')\n",
        "plt.title('Training loss')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 472
        },
        "id": "FxqdP-1bSF1j",
        "outputId": "6fb75729-8a43-4d58-d422-18121705026b"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#Generate characters \n",
        "#We must provide a sequence of seq_lenth as input to start the generation process\n",
        "#The prediction results is probabilities for each of the 103 characters at a specific\n",
        "#point in sequence. Let us pick the one with max probability and print it out.\n",
        "#Writing our own softmax function....\n",
        "def sample(preds):\n",
        "    preds = np.asarray(preds).astype('float64')\n",
        "    preds = np.log(preds)\n",
        "    exp_preds = np.exp(preds) #exp of log (x), isn't this same as x??\n",
        "    preds = exp_preds / np.sum(exp_preds)\n",
        "    probas = np.random.multinomial(1, preds, 1) \n",
        "    return np.argmax(probas)\n"
      ],
      "metadata": {
        "id": "QymUgjT-SQGC"
      },
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#Prediction\n",
        "#setting the seed\n",
        "# load the network weights\n",
        "filename = \"my_saved_weights_hindi_60epochs.h5\"\n",
        "model.load_weights(filename)\n",
        "\n",
        "#Pick a random sentence from the text as seed.\n",
        "start_index = random.randint(0, n_chars - seq_length - 1)\n",
        "\n",
        "#Initiate generated text and keep adding new predictions and print them out\n",
        "generated = ''\n",
        "sentence = raw_text[start_index: start_index + seq_length]\n",
        "generated += sentence\n",
        "\n",
        "print('----- Seed for our text prediction: \"' + sentence + '\"')\n",
        "#sys.stdout.write(generated)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bP31_r-SU89I",
        "outputId": "39482a37-515a-4e44-a04b-f6b743ba3b6d"
      },
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----- Seed for our text prediction: \"ै।\n",
            ".: अभ्यास में नहीं लगाए गए शास्त्रों के पाठ जहर हैं; भोजन\"\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# predicting the next 400 charecters\n",
        "for i in range(400):   # Number of characters including spaces\n",
        "    x_pred = np.zeros((1, seq_length, n_vocab))\n",
        "    for t, char in enumerate(sentence):\n",
        "        x_pred[0, t, char_to_int[char]] = 1.\n",
        "\n",
        "    preds = model.predict(x_pred, verbose=0)[0]\n",
        "    next_index = sample(preds)\n",
        "    next_char = int_to_char[next_index]\n",
        "\n",
        "    generated += next_char\n",
        "    sentence = sentence[1:] + next_char\n",
        "\n",
        "    sys.stdout.write(next_char)\n",
        "    sys.stdout.flush()\n",
        "print()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "A7nu2jwQVcjk",
        "outputId": "4eb55ebf-b623-4ca3-8c93-823fbe72691d"
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            " उसके लोगो सब में यदि बुदुष्य टाथ यह मरने वृ व्यक्ति  पाए एक एक बलत्रुण जो डपने पर्ता है, उसका ब्रभी भी निश मेश नोीं ौन|”\n",
            "“.:बादा रोज्यान प फवान रो कुम फर धन होते है|”\n",
            "“.:विसे सब बचयलग की मऔर मनिसे विल कर्ण से मालता है. यह भो वहीर हो जल मने परता है, उसकी अपने स्वबश्ति को होती और उसका मिव्य मभू धन उती भाए दिसबे सब गुव्या शरे को में मेन मनुश नहीं पढिया उसके अपनी प्तब राणान करे. ए. एक धिसी पर्बशते सी\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "#Stacked LSTM \n"
      ],
      "metadata": {
        "id": "2s_AGxwJWGtZ"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Deeper model woth 2 LSTM\n",
        "#To stack LSTM layers, we need to change the configuration of the prior \n",
        "#LSTM layer to output a 3D array as input for the subsequent layer.\n",
        "#We can do this by setting the return_sequences argument on the layer to True \n",
        "#(defaults to False). This will return one output for each input time step and provide a 3D array.\n",
        "#Below is the same example as above with return_sequences=True.\n",
        "\n",
        "model = Sequential()\n",
        "model.add(LSTM(128, input_shape=(seq_length, n_vocab), return_sequences=True))\n",
        "model.add(Dropout(0.2))\n",
        "model.add(LSTM(128))\n",
        "model.add(Dropout(0.2))\n",
        "model.add(Dense(n_vocab, activation='softmax'))\n",
        "\n",
        "optimizer = RMSprop(lr=0.01)\n",
        "model.compile(loss='categorical_crossentropy', optimizer=optimizer)\n",
        "model.summary()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yPusbjGdWOEm",
        "outputId": "23c774dc-d43c-4d12-b30b-fa73e0ae6cb8"
      },
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " lstm_1 (LSTM)               (None, 60, 128)           106496    \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 60, 128)           0         \n",
            "                                                                 \n",
            " lstm_2 (LSTM)               (None, 128)               131584    \n",
            "                                                                 \n",
            " dropout_1 (Dropout)         (None, 128)               0         \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 79)                10191     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 248,271\n",
            "Trainable params: 248,271\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# define the checkpoint\n",
        "from keras.callbacks import ModelCheckpoint\n",
        "\n",
        "filepath=\"saved_weights/saved_weights2-{epoch:02d}-{loss:.4f}.hdf5\"\n",
        "checkpoint = ModelCheckpoint(filepath, monitor='loss', verbose=1, save_best_only=True, mode='min')\n",
        "\n",
        "callbacks_list = [checkpoint]\n",
        "\n",
        "\n",
        "# Fit the model\n",
        "\n",
        "history2 = model.fit(x, y,\n",
        "          batch_size=128,\n",
        "          epochs=60,   \n",
        "          callbacks=callbacks_list)\n",
        "\n",
        "model.save('my_saved_weights_hindi2_60epochs.h5')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vFnix0S-WuLU",
        "outputId": "4ca46d7b-0b07-4301-e1cb-314809558e7c"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 3.6786\n",
            "Epoch 1: loss improved from inf to 3.67862, saving model to saved_weights/saved_weights2-01-3.6786.hdf5\n",
            "36/36 [==============================] - 4s 18ms/step - loss: 3.6786\n",
            "Epoch 2/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 3.2084\n",
            "Epoch 2: loss improved from 3.67862 to 3.20377, saving model to saved_weights/saved_weights2-02-3.2038.hdf5\n",
            "36/36 [==============================] - 1s 23ms/step - loss: 3.2038\n",
            "Epoch 3/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 2.7971\n",
            "Epoch 3: loss improved from 3.20377 to 2.79708, saving model to saved_weights/saved_weights2-03-2.7971.hdf5\n",
            "36/36 [==============================] - 1s 25ms/step - loss: 2.7971\n",
            "Epoch 4/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 2.5529\n",
            "Epoch 4: loss improved from 2.79708 to 2.55014, saving model to saved_weights/saved_weights2-04-2.5501.hdf5\n",
            "36/36 [==============================] - 1s 28ms/step - loss: 2.5501\n",
            "Epoch 5/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 2.3737\n",
            "Epoch 5: loss improved from 2.55014 to 2.36814, saving model to saved_weights/saved_weights2-05-2.3681.hdf5\n",
            "36/36 [==============================] - 1s 18ms/step - loss: 2.3681\n",
            "Epoch 6/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 2.2265\n",
            "Epoch 6: loss improved from 2.36814 to 2.22439, saving model to saved_weights/saved_weights2-06-2.2244.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 2.2244\n",
            "Epoch 7/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 2.0670\n",
            "Epoch 7: loss improved from 2.22439 to 2.06703, saving model to saved_weights/saved_weights2-07-2.0670.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 2.0670\n",
            "Epoch 8/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 1.9163\n",
            "Epoch 8: loss improved from 2.06703 to 1.92071, saving model to saved_weights/saved_weights2-08-1.9207.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 1.9207\n",
            "Epoch 9/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 1.7745\n",
            "Epoch 9: loss improved from 1.92071 to 1.77447, saving model to saved_weights/saved_weights2-09-1.7745.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 1.7745\n",
            "Epoch 10/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 1.6086\n",
            "Epoch 10: loss improved from 1.77447 to 1.60860, saving model to saved_weights/saved_weights2-10-1.6086.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 1.6086\n",
            "Epoch 11/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 1.4693\n",
            "Epoch 11: loss improved from 1.60860 to 1.47169, saving model to saved_weights/saved_weights2-11-1.4717.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 1.4717\n",
            "Epoch 12/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 1.3165\n",
            "Epoch 12: loss improved from 1.47169 to 1.31646, saving model to saved_weights/saved_weights2-12-1.3165.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 1.3165\n",
            "Epoch 13/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 1.1979\n",
            "Epoch 13: loss improved from 1.31646 to 1.19778, saving model to saved_weights/saved_weights2-13-1.1978.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 1.1978\n",
            "Epoch 14/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 1.0671\n",
            "Epoch 14: loss improved from 1.19778 to 1.06715, saving model to saved_weights/saved_weights2-14-1.0671.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 1.0671\n",
            "Epoch 15/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.9475\n",
            "Epoch 15: loss improved from 1.06715 to 0.94751, saving model to saved_weights/saved_weights2-15-0.9475.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.9475\n",
            "Epoch 16/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.8612\n",
            "Epoch 16: loss improved from 0.94751 to 0.86116, saving model to saved_weights/saved_weights2-16-0.8612.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.8612\n",
            "Epoch 17/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.7437\n",
            "Epoch 17: loss improved from 0.86116 to 0.74374, saving model to saved_weights/saved_weights2-17-0.7437.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.7437\n",
            "Epoch 18/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.6864\n",
            "Epoch 18: loss improved from 0.74374 to 0.68639, saving model to saved_weights/saved_weights2-18-0.6864.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.6864\n",
            "Epoch 19/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.6156\n",
            "Epoch 19: loss improved from 0.68639 to 0.61559, saving model to saved_weights/saved_weights2-19-0.6156.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.6156\n",
            "Epoch 20/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 0.5566\n",
            "Epoch 20: loss improved from 0.61559 to 0.55827, saving model to saved_weights/saved_weights2-20-0.5583.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.5583\n",
            "Epoch 21/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.5450\n",
            "Epoch 21: loss improved from 0.55827 to 0.54501, saving model to saved_weights/saved_weights2-21-0.5450.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.5450\n",
            "Epoch 22/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.4746\n",
            "Epoch 22: loss improved from 0.54501 to 0.47390, saving model to saved_weights/saved_weights2-22-0.4739.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.4739\n",
            "Epoch 23/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.4619\n",
            "Epoch 23: loss improved from 0.47390 to 0.46194, saving model to saved_weights/saved_weights2-23-0.4619.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.4619\n",
            "Epoch 24/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.4046\n",
            "Epoch 24: loss improved from 0.46194 to 0.40457, saving model to saved_weights/saved_weights2-24-0.4046.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.4046\n",
            "Epoch 25/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.3891\n",
            "Epoch 25: loss improved from 0.40457 to 0.38906, saving model to saved_weights/saved_weights2-25-0.3891.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.3891\n",
            "Epoch 26/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.3655\n",
            "Epoch 26: loss improved from 0.38906 to 0.36550, saving model to saved_weights/saved_weights2-26-0.3655.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.3655\n",
            "Epoch 27/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.3552\n",
            "Epoch 27: loss improved from 0.36550 to 0.35798, saving model to saved_weights/saved_weights2-27-0.3580.hdf5\n",
            "36/36 [==============================] - 1s 16ms/step - loss: 0.3580\n",
            "Epoch 28/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.3426\n",
            "Epoch 28: loss improved from 0.35798 to 0.34350, saving model to saved_weights/saved_weights2-28-0.3435.hdf5\n",
            "36/36 [==============================] - 1s 14ms/step - loss: 0.3435\n",
            "Epoch 29/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.3382\n",
            "Epoch 29: loss improved from 0.34350 to 0.33925, saving model to saved_weights/saved_weights2-29-0.3392.hdf5\n",
            "36/36 [==============================] - 1s 16ms/step - loss: 0.3392\n",
            "Epoch 30/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.3022\n",
            "Epoch 30: loss improved from 0.33925 to 0.30665, saving model to saved_weights/saved_weights2-30-0.3066.hdf5\n",
            "36/36 [==============================] - 1s 16ms/step - loss: 0.3066\n",
            "Epoch 31/60\n",
            "32/36 [=========================>....] - ETA: 0s - loss: 0.2999\n",
            "Epoch 31: loss did not improve from 0.30665\n",
            "36/36 [==============================] - 1s 14ms/step - loss: 0.3085\n",
            "Epoch 32/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.2887\n",
            "Epoch 32: loss improved from 0.30665 to 0.29444, saving model to saved_weights/saved_weights2-32-0.2944.hdf5\n",
            "36/36 [==============================] - 0s 13ms/step - loss: 0.2944\n",
            "Epoch 33/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.2992\n",
            "Epoch 33: loss did not improve from 0.29444\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2998\n",
            "Epoch 34/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2740\n",
            "Epoch 34: loss improved from 0.29444 to 0.27397, saving model to saved_weights/saved_weights2-34-0.2740.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.2740\n",
            "Epoch 35/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.2767\n",
            "Epoch 35: loss did not improve from 0.27397\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2782\n",
            "Epoch 36/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.2843\n",
            "Epoch 36: loss did not improve from 0.27397\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2831\n",
            "Epoch 37/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2782\n",
            "Epoch 37: loss did not improve from 0.27397\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2782\n",
            "Epoch 38/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2793\n",
            "Epoch 38: loss did not improve from 0.27397\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2793\n",
            "Epoch 39/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.2374\n",
            "Epoch 39: loss improved from 0.27397 to 0.23746, saving model to saved_weights/saved_weights2-39-0.2375.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.2375\n",
            "Epoch 40/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.2601\n",
            "Epoch 40: loss did not improve from 0.23746\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.2601\n",
            "Epoch 41/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.2491\n",
            "Epoch 41: loss did not improve from 0.23746\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2519\n",
            "Epoch 42/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2407\n",
            "Epoch 42: loss did not improve from 0.23746\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2407\n",
            "Epoch 43/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 0.2556\n",
            "Epoch 43: loss did not improve from 0.23746\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.2542\n",
            "Epoch 44/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2565\n",
            "Epoch 44: loss did not improve from 0.23746\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2565\n",
            "Epoch 45/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2307\n",
            "Epoch 45: loss improved from 0.23746 to 0.23068, saving model to saved_weights/saved_weights2-45-0.2307.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.2307\n",
            "Epoch 46/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2254\n",
            "Epoch 46: loss improved from 0.23068 to 0.22540, saving model to saved_weights/saved_weights2-46-0.2254.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.2254\n",
            "Epoch 47/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.2351\n",
            "Epoch 47: loss did not improve from 0.22540\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2367\n",
            "Epoch 48/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2453\n",
            "Epoch 48: loss did not improve from 0.22540\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2453\n",
            "Epoch 49/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2211\n",
            "Epoch 49: loss improved from 0.22540 to 0.22108, saving model to saved_weights/saved_weights2-49-0.2211.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.2211\n",
            "Epoch 50/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2296\n",
            "Epoch 50: loss did not improve from 0.22108\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2296\n",
            "Epoch 51/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2246\n",
            "Epoch 51: loss did not improve from 0.22108\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2246\n",
            "Epoch 52/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2180\n",
            "Epoch 52: loss improved from 0.22108 to 0.21804, saving model to saved_weights/saved_weights2-52-0.2180.hdf5\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.2180\n",
            "Epoch 53/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 0.2085\n",
            "Epoch 53: loss improved from 0.21804 to 0.21009, saving model to saved_weights/saved_weights2-53-0.2101.hdf5\n",
            "36/36 [==============================] - 0s 13ms/step - loss: 0.2101\n",
            "Epoch 54/60\n",
            "36/36 [==============================] - ETA: 0s - loss: 0.2111\n",
            "Epoch 54: loss did not improve from 0.21009\n",
            "36/36 [==============================] - 0s 11ms/step - loss: 0.2111\n",
            "Epoch 55/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.2061\n",
            "Epoch 55: loss improved from 0.21009 to 0.20624, saving model to saved_weights/saved_weights2-55-0.2062.hdf5\n",
            "36/36 [==============================] - 0s 14ms/step - loss: 0.2062\n",
            "Epoch 56/60\n",
            "35/36 [============================>.] - ETA: 0s - loss: 0.2172\n",
            "Epoch 56: loss did not improve from 0.20624\n",
            "36/36 [==============================] - 0s 12ms/step - loss: 0.2162\n",
            "Epoch 57/60\n",
            "34/36 [===========================>..] - ETA: 0s - loss: 0.2138\n",
            "Epoch 57: loss did not improve from 0.20624\n",
            "36/36 [==============================] - 0s 13ms/step - loss: 0.2165\n",
            "Epoch 58/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.1918\n",
            "Epoch 58: loss improved from 0.20624 to 0.19660, saving model to saved_weights/saved_weights2-58-0.1966.hdf5\n",
            "36/36 [==============================] - 0s 13ms/step - loss: 0.1966\n",
            "Epoch 59/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.1985\n",
            "Epoch 59: loss did not improve from 0.19660\n",
            "36/36 [==============================] - 0s 13ms/step - loss: 0.2024\n",
            "Epoch 60/60\n",
            "33/36 [==========================>...] - ETA: 0s - loss: 0.2007\n",
            "Epoch 60: loss did not improve from 0.19660\n",
            "36/36 [==============================] - 1s 14ms/step - loss: 0.2020\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#Tracking loos function \n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# Get the loss values from the training history\n",
        "loss = history2.history['loss']\n",
        "\n",
        "# Create a plot of the loss over epochs\n",
        "plt.plot(loss)\n",
        "plt.title('Model Loss')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Loss')\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 472
        },
        "id": "oG9v0J9YXH1f",
        "outputId": "e220100c-2be4-4133-c4cd-c8c4c452c552"
      },
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#Generate characters \n",
        "#We must provide a sequence of seq_lenth as input to start the generation process\n",
        "#The prediction results is probabilities for each of the 103 characters at a specific\n",
        "#point in sequence. Let us pick the one with max probability and print it out.\n",
        "#Writing our own softmax function....\n",
        "def sample(preds):\n",
        "    preds = np.asarray(preds).astype('float64')\n",
        "    preds = np.log(preds)\n",
        "    exp_preds = np.exp(preds) #exp of log (x), isn't this same as x??\n",
        "    preds = exp_preds / np.sum(exp_preds)\n",
        "    probas = np.random.multinomial(1, preds, 1) \n",
        "    return np.argmax(probas)\n"
      ],
      "metadata": {
        "id": "akzSdHx4Z17l"
      },
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#Prediction\n",
        "#setting the seed\n",
        "# load the network weights\n",
        "filename = \"my_saved_weights_hindi2_60epochs.h5\"\n",
        "model.load_weights(filename)\n",
        "\n",
        "#Pick a random sentence from the text as seed.\n",
        "start_index = random.randint(0, n_chars - seq_length - 1)\n",
        "\n",
        "#Initiate generated text and keep adding new predictions and print them out\n",
        "generated = ''\n",
        "sentence = raw_text[start_index: start_index + seq_length]\n",
        "generated += sentence\n",
        "\n",
        "print('----- Seed for our text prediction: \"' + sentence + '\"')\n",
        "#sys.stdout.write(generated)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "loo-F1XAaAe9",
        "outputId": "c2596a47-09cb-45e1-aaa2-d246c636621e"
      },
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----- Seed for our text prediction: \" माता है. अध्यात्मिक ज्ञान मेरा पिता है. धर्माचरण मेरा बंधू \"\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# predicting the next 400 charecters\n",
        "for i in range(400):   # Number of characters including spaces\n",
        "    x_pred = np.zeros((1, seq_length, n_vocab))\n",
        "    for t, char in enumerate(sentence):\n",
        "        x_pred[0, t, char_to_int[char]] = 1.\n",
        "\n",
        "    preds = model.predict(x_pred, verbose=0)[0]\n",
        "    next_index = sample(preds)\n",
        "    next_char = int_to_char[next_index]\n",
        "\n",
        "    generated += next_char\n",
        "    sentence = sentence[1:] + next_char\n",
        "\n",
        "    sys.stdout.write(next_char)\n",
        "    sys.stdout.flush()\n",
        "print()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9z8CmfHUaLGT",
        "outputId": "10d306d7-d1d2-4aa7-c522-66cbbda52b5a"
      },
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "है. वही सुव्वव्म्रे को लोगा की अपनी आत्वति से में तक उसका लाण वन्ह्रण को ता प\n",
            "रे समला है. धर सोश बदल जो समयक गन कररे वाले वाली के भाजन समें आप उनकी उसकर मिलकत जब फदट भी शरी से माधन से उसकूर वह शरण जो चालण जो मिमेष है की बार इसही अत्योिशि को प्रत्रि को सहैववही से प्रद्यि है.इल की अपनी भोजी को अपनी स्वच्कि से छाश मानि को अपनी प्राप्ति को भी फी शत्ति सा पुण वह एक ला पाआज आपक्रा भी भदी है, मे सकता हो \n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# Single LSTM model loss\n",
        "loss_single_lstm = history.history['loss']\n",
        "epochs_single_lstm = range(1, len(loss_single_lstm) + 1)\n",
        "\n",
        "# Stacked LSTM model loss\n",
        "loss_stacked_lstm = history2.history['loss']\n",
        "epochs_stacked_lstm = range(1, len(loss_stacked_lstm) + 1)\n",
        "\n",
        "# Create a plot of the loss for both models\n",
        "plt.plot(epochs_single_lstm, loss_single_lstm, 'y', label='Single LSTM')\n",
        "plt.plot(epochs_stacked_lstm, loss_stacked_lstm, 'b', label='Stacked LSTM')\n",
        "plt.title('Model Loss Comparison')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 472
        },
        "id": "MGjTEJx6Zddo",
        "outputId": "819c2221-7898-43a3-c132-7b06d6f80a4e"
      },
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    }
  ]
}