analysis.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import bz2\n",
    "from tqdm import tqdm\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import nltk\n",
    "import os\n",
    "from kaggle.api.kaggle_api_extended import KaggleApi\n",
    "from dotenv import load_dotenv\n",
    "\n",
    "load_dotenv() "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Loading"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set your Kaggle API credentials\n",
    "os.environ[\"KAGGLE_USERNAME\"] = os.getenv(\"KAGGLE_USERNAME\")\n",
    "os.environ[\"KAGGLE_KEY\"] = os.getenv(\"KAGGLE_KEY\")\n",
    "\n",
    "# Instantiate the Kaggle API client\n",
    "api = KaggleApi()\n",
    "api.authenticate()\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset URL: https://www.kaggle.com/datasets/bittlingmayer/amazonreviews\n"
     ]
    }
   ],
   "source": [
    "# Name of the dataset you want to download\n",
    "dataset_name = \"bittlingmayer/amazonreviews\"\n",
    "\n",
    "# Directory to download the dataset\n",
    "download_dir = \"./data\"\n",
    "\n",
    "# Create the directory if it doesn't exist\n",
    "os.makedirs(download_dir, exist_ok=True)\n",
    "\n",
    "# Download the dataset\n",
    "api.dataset_download_files(dataset_name, path=download_dir, unzip=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Successful \n",
      "Train Successful \n"
     ]
    }
   ],
   "source": [
    "output_path='output/test.ft.txt'\n",
    "with bz2.open('data/test.ft.txt.bz2', 'rt', encoding='utf-8') as compressed_file, open(output_path, 'w', encoding='utf-8') as output_file:\n",
    "    for line in compressed_file:\n",
    "        output_file.write(line)\n",
    "    print('Test Successful ')\n",
    "\n",
    "output_path='output/train.ft.txt'\n",
    "with bz2.open('data/train.ft.txt.bz2', 'rt', encoding='utf-8') as compressed_file, open(output_path, 'w', encoding='utf-8') as output_file:\n",
    "    for line in compressed_file:\n",
    "        output_file.write(line)\n",
    "    print('Train Successful ')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 3600000/3600000 [00:03<00:00, 916176.14it/s] \n",
      "100%|██████████| 400000/400000 [00:00<00:00, 863328.39it/s]\n"
     ]
    }
   ],
   "source": [
    "train,test,train_label,test_label=[],[],[],[]\n",
    "with open('output/train.ft.txt', 'r', encoding='utf-8') as file:\n",
    "    lines = file.readlines()\n",
    "for line in tqdm(lines):\n",
    "    train.append(line.split('__label__')[1][1:])\n",
    "    train_label.append(line.split('__label__')[1][0])\n",
    "with open('output/test.ft.txt', 'r', encoding='utf-8') as file:\n",
    "    lines = file.readlines()\n",
    "for line in tqdm(lines):\n",
    "    test.append(line.split('__label__')[1][1:])\n",
    "    test_label.append(line.split('__label__')[1][0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "' Stuning even for the non-gamer: This sound track was beautiful! It paints the senery in your mind so well I would recomend it even to people who hate vid. game music! I have played the game Chrono Cross but out of all of the games I have ever played it has the best music! It backs away from crude keyboarding and takes a fresher step with grate guitars and soulful orchestras. It would impress anyone who cares to listen! ^_^\\n'"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Now, let's import this data into pandas DataFrames\n",
    "train_df = pd.DataFrame({'text': train, 'label': train_label})\n",
    "test_df = pd.DataFrame({'text': test, 'label': test_label})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'2'"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df['label'][0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## PreProcess"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package punkt to\n",
      "[nltk_data]     C:\\Users\\faiza\\AppData\\Roaming\\nltk_data...\n",
      "[nltk_data]   Package punkt is already up-to-date!\n",
      "[nltk_data] Downloading package stopwords to\n",
      "[nltk_data]     C:\\Users\\faiza\\AppData\\Roaming\\nltk_data...\n",
      "[nltk_data]   Package stopwords is already up-to-date!\n",
      "[nltk_data] Downloading package wordnet to\n",
      "[nltk_data]     C:\\Users\\faiza\\AppData\\Roaming\\nltk_data...\n",
      "[nltk_data]   Package wordnet is already up-to-date!\n"
     ]
    }
   ],
   "source": [
    "import nltk\n",
    "from nltk.corpus import stopwords\n",
    "from nltk.tokenize import word_tokenize\n",
    "from nltk.stem import WordNetLemmatizer\n",
    "import string\n",
    "\n",
    "# Download NLTK resources if not already downloaded\n",
    "nltk.download('punkt')\n",
    "nltk.download('stopwords')\n",
    "nltk.download('wordnet')\n",
    "\n",
    "# Initialize WordNet Lemmatizer\n",
    "lemmatizer = WordNetLemmatizer()\n",
    "\n",
    "# Preprocessing function\n",
    "def preprocess_text(text):\n",
    "    # Tokenization\n",
    "    tokens = word_tokenize(text)\n",
    "    \n",
    "    # Lowercasing and removing punctuation\n",
    "    tokens = [word.lower() for word in tokens if word.isalnum()]\n",
    "    \n",
    "    # Removing stopwords\n",
    "    stop_words = set(stopwords.words('english'))\n",
    "    tokens = [word for word in tokens if word not in stop_words]\n",
    "    \n",
    "    # Lemmatization\n",
    "    tokens = [lemmatizer.lemmatize(word) for word in tokens]\n",
    "    \n",
    "    return tokens\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Stuning even for the non-gamer: This sound track was beautiful! It paints the senery in your mind so well I would recomend it even to people who hate vid. game music! I have played the game Chrono Cross but out of all of the games I have ever played it has the best music! It backs away from crude keyboarding and takes a fresher step with grate guitars and soulful orchestras. It would impress anyone who cares to listen! ^_^\n",
      "\n"
     ]
    }
   ],
   "source": [
    "example=train_df['text'][0]\n",
    "print(example)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Stuning',\n",
       " 'even',\n",
       " 'for',\n",
       " 'the',\n",
       " 'non-gamer',\n",
       " ':',\n",
       " 'This',\n",
       " 'sound',\n",
       " 'track',\n",
       " 'was']"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tokens = nltk.word_tokenize(example)\n",
    "tokens[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['stuning',\n",
       " 'even',\n",
       " 'sound',\n",
       " 'track',\n",
       " 'beautiful',\n",
       " 'paint',\n",
       " 'senery',\n",
       " 'mind',\n",
       " 'well',\n",
       " 'would']"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocess_text(example)[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'neg': 0.088, 'neu': 0.669, 'pos': 0.243, 'compound': 0.9454}\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package vader_lexicon to\n",
      "[nltk_data]     C:\\Users\\faiza\\AppData\\Roaming\\nltk_data...\n",
      "[nltk_data]   Package vader_lexicon is already up-to-date!\n"
     ]
    }
   ],
   "source": [
    "import nltk\n",
    "from nltk.sentiment.vader import SentimentIntensityAnalyzer\n",
    "\n",
    "# Download NLTK resources if not already downloaded\n",
    "nltk.download('vader_lexicon')\n",
    "\n",
    "# Initialize SentimentIntensityAnalyzer\n",
    "sid = SentimentIntensityAnalyzer()\n",
    "\n",
    "# Example usage\n",
    "sentiment_scores = sid.polarity_scores(example)\n",
    "print(sentiment_scores)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## EDA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3600000, 2)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['2', '1'], dtype=object)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df['label'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x300 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "label_counts = train_df['label'].value_counts(normalize=True) * 100\n",
    "\n",
    "# Define labels for the pie chart\n",
    "labels = ['Negative (1)', 'Positive (2)']\n",
    "\n",
    "# Plotting\n",
    "plt.figure(figsize=(8, 3))\n",
    "plt.pie(label_counts, labels=labels, autopct='%1.1f%%', colors=['skyblue', 'lightgreen'], startangle=140)\n",
    "plt.title('Percentage of Train Each Label')\n",
    "plt.axis('equal') \n",
    "\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x300 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "label_counts = test_df['label'].value_counts(normalize=True) * 100\n",
    "\n",
    "# Define labels for the pie chart\n",
    "labels = ['Negative (1)', 'Positive (2)']\n",
    "\n",
    "# Plotting\n",
    "plt.figure(figsize=(8, 3))\n",
    "plt.pie(label_counts, labels=labels, autopct='%1.1f%%', colors=['skyblue', 'lightgreen'], startangle=140)\n",
    "plt.title('Percentage of Negative Each Label')\n",
    "plt.axis('equal') \n",
    "\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### note\n",
    " 1. Data is Balanced"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import AutoTokenizer\n",
    "from transformers import AutoModelForSequenceClassification\n",
    "from scipy.special import softmax"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "MODEL = f\"cardiffnlp/twitter-roberta-base-sentiment\"\n",
    "tokenizer = AutoTokenizer.from_pretrained(MODEL)\n",
    "model = AutoModelForSequenceClassification.from_pretrained(MODEL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Stuning even for the non-gamer: This sound track was beautiful! It paints the senery in your mind so well I would recomend it even to people who hate vid. game music! I have played the game Chrono Cross but out of all of the games I have ever played it has the best music! It backs away from crude keyboarding and takes a fresher step with grate guitars and soulful orchestras. It would impress anyone who cares to listen! ^_^\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'neg': 0.088, 'neu': 0.669, 'pos': 0.243, 'compound': 0.9454}"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# VADER results on example\n",
    "print(example)\n",
    "sid.polarity_scores(example)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'roberta_neg': 0.0006018789, 'roberta_neu': 0.0030538081, 'roberta_pos': 0.99634427}\n"
     ]
    }
   ],
   "source": [
    "# Run for Roberta Model\n",
    "encoded_text = tokenizer(example, return_tensors='pt')\n",
    "output = model(**encoded_text)\n",
    "scores = output[0][0].detach().numpy()\n",
    "scores = softmax(scores)\n",
    "scores_dict = {\n",
    "    'roberta_neg' : scores[0],\n",
    "    'roberta_neu' : scores[1],\n",
    "    'roberta_pos' : scores[2]\n",
    "}\n",
    "print(scores_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SequenceClassifierOutput(loss=None, logits=tensor([[-2.7554, -1.1313,  4.6564]], grad_fn=<AddmmBackward0>), hidden_states=None, attentions=None)"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "output"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Note\n",
    "1. Sid doesnot have understanding relationship and context\n",
    "2. roberta have  have understanding relationship and context"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Stuning even for the non-gamer: This sound track was beautiful! It paints the senery in your mind so well I would recomend it even to people who hate vid. game music! I have played the game Chrono Cross but out of all of the games I have ever played it has the best music! It backs away from crude keyboarding and takes a fresher step with grate guitars and soulful orchestras. It would impress anyone who cares to listen! ^_^\n",
      " 2\n"
     ]
    }
   ],
   "source": [
    "print(train_df['text'][0],\n",
    "train_df['label'][0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import tok"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import Trainer, TrainingArguments\n",
    "from sklearn.metrics import accuracy_score\n",
    "import torch\n",
    "from torch.utils.data import Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RobertaTokenizerFast(name_or_path='cardiffnlp/twitter-roberta-base-sentiment', vocab_size=50265, model_max_length=1000000000000000019884624838656, is_fast=True, padding_side='right', truncation_side='right', special_tokens={'bos_token': '<s>', 'eos_token': '</s>', 'unk_token': '<unk>', 'sep_token': '</s>', 'pad_token': '<pad>', 'cls_token': '<s>', 'mask_token': '<mask>'}, clean_up_tokenization_spaces=True),  added_tokens_decoder={\n",
       "\t0: AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True, special=True),\n",
       "\t1: AddedToken(\"<pad>\", rstrip=False, lstrip=False, single_word=False, normalized=True, special=True),\n",
       "\t2: AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True, special=True),\n",
       "\t3: AddedToken(\"<unk>\", rstrip=False, lstrip=False, single_word=False, normalized=True, special=True),\n",
       "\t50264: AddedToken(\"<mask>\", rstrip=False, lstrip=True, single_word=False, normalized=False, special=True),\n",
       "}"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tokenizer"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Taking Small Dataset for Training "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "# taking small dataset for training\n",
    "train_text = train_df['text'][0:100]\n",
    "train_label = train_df['label'][0:100]\n",
    "test_text = train_df['text'][0:10]\n",
    "test_label = train_df['label'][0:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "' Caution!: These tracks are not the \"original\" versions but are re-recorded versions. So, whether the tracks are \"remastered\" or not is irrelevant.\\n'"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_text[99]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "class CustomDataset(Dataset):\n",
    "    def __init__(self, texts, labels, tokenizer, max_length):\n",
    "        self.texts = texts\n",
    "        self.labels = labels\n",
    "        self.tokenizer = tokenizer\n",
    "        self.max_length = max_length\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.texts)\n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        text = str(self.texts[idx])\n",
    "        label = int(self.labels[idx])\n",
    "\n",
    "        encoding = self.tokenizer.encode_plus(\n",
    "            text,\n",
    "            add_special_tokens=True,\n",
    "            max_length=self.max_length,\n",
    "            return_token_type_ids=False,\n",
    "            padding=\"max_length\",\n",
    "            truncation=True,\n",
    "            return_attention_mask=True,\n",
    "            return_tensors='pt',\n",
    "        )\n",
    "\n",
    "        return {\n",
    "            'input_ids': encoding['input_ids'].flatten(),\n",
    "            'attention_mask': encoding['attention_mask'].flatten(),\n",
    "            'labels': torch.tensor(label, dtype=torch.long)\n",
    "        }\n",
    "\n",
    "train_dataset = CustomDataset(train_text, train_label, tokenizer, max_length=128)\n",
    "test_dataset = CustomDataset(test_text, test_label, tokenizer, max_length=128)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\faiza\\Music\\tut\\nlp\\sentimate\\env\\Lib\\site-packages\\accelerate\\accelerator.py:436: FutureWarning: Passing the following arguments to `Accelerator` is deprecated and will be removed in version 1.0 of Accelerate: dict_keys(['dispatch_batches', 'split_batches', 'even_batches', 'use_seedable_sampler']). Please pass an `accelerate.DataLoaderConfiguration` instead: \n",
      "dataloader_config = DataLoaderConfiguration(dispatch_batches=None, split_batches=False, even_batches=True, use_seedable_sampler=True)\n",
      "  warnings.warn(\n"
     ]
    }
   ],
   "source": [
    "# Define training arguments\n",
    "training_args = TrainingArguments(\n",
    "    output_dir=\"./results\",\n",
    "    num_train_epochs=3,\n",
    "    per_device_train_batch_size=8,\n",
    "    per_device_eval_batch_size=8,\n",
    "    warmup_steps=500,\n",
    "    weight_decay=0.01,\n",
    "    logging_dir=\"./logs\",\n",
    ")\n",
    "\n",
    "# Define the Trainer\n",
    "trainer = Trainer(\n",
    "    model=model,\n",
    "    args=training_args,\n",
    "    train_dataset=train_dataset,\n",
    "    eval_dataset=test_dataset,\n",
    "    compute_metrics=lambda pred: {\"accuracy\": accuracy_score(pred.label_ids, pred.predictions.argmax(-1))},\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 14/1350000 [09:44<15657:27:36, 41.75s/it]\n",
      " 13%|█▎        | 5/39 [02:12<15:01, 26.51s/it]\n",
      "100%|██████████| 39/39 [05:18<00:00,  8.16s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'train_runtime': 318.3522, 'train_samples_per_second': 0.942, 'train_steps_per_second': 0.123, 'train_loss': 0.26884976411477113, 'epoch': 3.0}\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "TrainOutput(global_step=39, training_loss=0.26884976411477113, metrics={'train_runtime': 318.3522, 'train_samples_per_second': 0.942, 'train_steps_per_second': 0.123, 'train_loss': 0.26884976411477113, 'epoch': 3.0})"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trainer.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('./fine_tuned_model\\\\tokenizer_config.json',\n",
       " './fine_tuned_model\\\\special_tokens_map.json',\n",
       " './fine_tuned_model\\\\vocab.json',\n",
       " './fine_tuned_model\\\\merges.txt',\n",
       " './fine_tuned_model\\\\added_tokens.json',\n",
       " './fine_tuned_model\\\\tokenizer.json')"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Save the fine-tuned model\n",
    "output_model_dir = \"./fine_tuned_model\"\n",
    "model.save_pretrained(output_model_dir)\n",
    "tokenizer.save_pretrained(output_model_dir)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load the model FINE TUNE MODEL"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import AutoTokenizer, AutoModelForSequenceClassification\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'Negative': 0.005059136, 'Netrual': 0.7918837, 'Postive': 0.20305713}\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# Load the fine-tuned model and tokenizer\n",
    "model_dir = \"./fine_tuned_model\"\n",
    "tokenizer = AutoTokenizer.from_pretrained(model_dir)\n",
    "model = AutoModelForSequenceClassification.from_pretrained(model_dir)\n",
    "\n",
    "# Example text for prediction\n",
    "text = \"This is a packing is standard!\"\n",
    "\n",
    "# Tokenize the text\n",
    "inputs = tokenizer(text, return_tensors=\"pt\")\n",
    "\n",
    "# Perform prediction\n",
    "output = model(**inputs)\n",
    "scores = output[0][0].detach().numpy()\n",
    "scores = softmax(scores)\n",
    "scores_dict = {\n",
    "    'Negative' : scores[0],\n",
    "    'Netrual' : scores[1],\n",
    "    'Postive' : scores[2]\n",
    "}\n",
    "print(scores_dict)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Maximum value: 0.7918837\n",
      "Key with maximum value: Netrual\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# Find the key corresponding to the maximum value\n",
    "max_key = max(scores_dict, key=scores_dict.get)\n",
    "\n",
    "# Get the maximum value\n",
    "max_value = scores_dict[max_key]\n",
    "\n",
    "print(\"Maximum value:\", max_value)\n",
    "print(\"Key with maximum value:\", max_key)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Push your Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1c1ceeff09d74c949616fc86307245d1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "VBox(children=(HTML(value='<center> <img\\nsrc=https://huggingface.co/front/assets/huggingface_logo-noborder.sv…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from huggingface_hub import notebook_login\n",
    "\n",
    "notebook_login()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "42e950490c0841b293ee3370bcc4a485",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "model.safetensors:   0%|          | 0.00/499M [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "CommitInfo(commit_url='https://huggingface.co/faizack/Fine_Tune_Roberta-base-sentiment/commit/e5a79077207633adb95215295c1bd173449bf296', commit_message='Upload RobertaForSequenceClassification', commit_description='', oid='e5a79077207633adb95215295c1bd173449bf296', pr_url=None, pr_revision=None, pr_num=None)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.push_to_hub(\"faizack/Fine_Tune_Roberta-base-sentiment\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "env",
   "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": "3.12.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}