bert-relevance.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Unique values in labels after cleaning: [0 1]\n",
      "Train data size: 3047 Test data size: 762\n",
      "Sample training labels: 3643    0\n",
      "2981    1\n",
      "1351    0\n",
      "3418    0\n",
      "246     0\n",
      "Name: V9_KKI_Relevanz \\n0 = nicht relevant, 1 = relevant, dtype: int32\n",
      "Sample testing labels: 1011    1\n",
      "3554    1\n",
      "3697    1\n",
      "897     0\n",
      "3244    0\n",
      "Name: V9_KKI_Relevanz \\n0 = nicht relevant, 1 = relevant, dtype: int32\n",
      "Training set label distribution:\n",
      " V9_KKI_Relevanz \\n0 = nicht relevant, 1 = relevant\n",
      "0    2124\n",
      "1     923\n",
      "Name: count, dtype: int64\n",
      "Test set label distribution:\n",
      " V9_KKI_Relevanz \\n0 = nicht relevant, 1 = relevant\n",
      "0    558\n",
      "1    204\n",
      "Name: count, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# Load the data\n",
    "data = pd.read_csv('media_test_100.csv')  # Update the path if necessary\n",
    "\n",
    "# Select relevant columns\n",
    "texts = data['Artikeltext_Auto']\n",
    "labels = data['V9_KKI_Relevanz \\n0 = nicht relevant, 1 = relevant']\n",
    "\n",
    "# Drop rows with NaN labels\n",
    "filtered_data = data.dropna(subset=['V9_KKI_Relevanz \\n0 = nicht relevant, 1 = relevant'])\n",
    "\n",
    "# Extract texts and labels again after filtering out NaNs\n",
    "texts = filtered_data['Artikeltext_Auto'].reset_index(drop=True)\n",
    "labels = filtered_data['V9_KKI_Relevanz \\n0 = nicht relevant, 1 = relevant'].reset_index(drop=True)\n",
    "\n",
    "# Convert labels to integers (0 or 1)\n",
    "labels = labels.astype(int)\n",
    "\n",
    "# Print out the unique values in labels to verify they are 0 and 1\n",
    "print(\"Unique values in labels after cleaning:\", labels.unique())\n",
    "\n",
    "# Train-test split (reproducible split with a fixed random state)\n",
    "train_texts, test_texts, train_labels, test_labels = train_test_split(\n",
    "    texts, labels, test_size=0.2, random_state=42\n",
    ")\n",
    "\n",
    "# Convert labels to integers\n",
    "train_labels = train_labels.astype(int)\n",
    "test_labels = test_labels.astype(int)\n",
    "\n",
    "# Print sizes and a few examples of the split data for verification\n",
    "print('Train data size:', len(train_labels), 'Test data size:', len(test_labels))\n",
    "print('Sample training labels:', train_labels.head())\n",
    "print('Sample testing labels:', test_labels.head())\n",
    "\n",
    "# Check the distribution of labels in training and test sets\n",
    "print(\"Training set label distribution:\\n\", train_labels.value_counts())\n",
    "print(\"Test set label distribution:\\n\", test_labels.value_counts())\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import AutoTokenizer\n",
    "from datasets import Dataset\n",
    "import pandas as pd\n",
    "\n",
    "# Function to prepare and tokenize datasets based on the given model and tokenizer\n",
    "def prepare_dataset(train_texts, test_texts, train_labels, test_labels, model_id):\n",
    "    # Load the tokenizer specific to the model\n",
    "    tokenizer = AutoTokenizer.from_pretrained(model_id)\n",
    "\n",
    "    # Tokenize function\n",
    "    def tokenize_function(examples):\n",
    "        return tokenizer(examples['text'], padding=\"max_length\", truncation=True)\n",
    "\n",
    "    # Prepare the datasets\n",
    "    train_dataset = Dataset.from_pandas(pd.DataFrame({'text': train_texts, 'label': train_labels}))\n",
    "    test_dataset = Dataset.from_pandas(pd.DataFrame({'text': test_texts, 'label': test_labels}))\n",
    "\n",
    "    # Verify the initial state of datasets\n",
    "    print(\"Initial train dataset sample:\")\n",
    "    print(train_dataset[0])\n",
    "    print(\"Initial test dataset sample:\")\n",
    "    print(test_dataset[0])\n",
    "\n",
    "    # Tokenize the datasets\n",
    "    train_dataset = train_dataset.map(tokenize_function, batched=True)\n",
    "    test_dataset = test_dataset.map(tokenize_function, batched=True)\n",
    "\n",
    "    # Rename the label column to 'labels'\n",
    "    train_dataset = train_dataset.rename_column(\"label\", \"labels\")\n",
    "    test_dataset = test_dataset.rename_column(\"label\", \"labels\")\n",
    "\n",
    "    # Print a sample after tokenization\n",
    "    print(\"\\nTokenized train dataset sample:\")\n",
    "    print(train_dataset[0])\n",
    "    print(\"Tokenized test dataset sample:\")\n",
    "    print(test_dataset[0])\n",
    "\n",
    "    # Set format for PyTorch\n",
    "    train_dataset.set_format(type=\"torch\", columns=[\"input_ids\", \"attention_mask\", \"labels\"])\n",
    "    test_dataset.set_format(type=\"torch\", columns=[\"input_ids\", \"attention_mask\", \"labels\"])\n",
    "\n",
    "    # Verify that the labels are correctly included and formatted\n",
    "    print(\"\\nFinal train dataset sample after format:\")\n",
    "    print(train_dataset[0])\n",
    "    print(\"Final test dataset sample after format:\")\n",
    "    print(test_dataset[0])\n",
    "\n",
    "    return train_dataset, test_dataset\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import AutoModelForSequenceClassification, TrainingArguments, Trainer, EarlyStoppingCallback\n",
    "import numpy as np\n",
    "from sklearn.metrics import accuracy_score, precision_recall_fscore_support\n",
    "import torch\n",
    "\n",
    "# Define a reusable function for model training and evaluation with early stopping\n",
    "def train_and_evaluate(model_id):\n",
    "    # Prepare the datasets\n",
    "    train_dataset, test_dataset = prepare_dataset(train_texts, test_texts, train_labels, test_labels, model_id)\n",
    "\n",
    "    # Load the model specific to the model_id\n",
    "    model = AutoModelForSequenceClassification.from_pretrained(\n",
    "        model_id, \n",
    "        num_labels=2, \n",
    "        id2label={0: \"not relevant\", 1: \"relevant\"}, \n",
    "        label2id={\"not relevant\": 0, \"relevant\": 1}\n",
    "    )\n",
    "    device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "    model.to(device)\n",
    "\n",
    "    # Define the training arguments with early stopping enabled\n",
    "    training_args = TrainingArguments(\n",
    "        output_dir=f\"{model_id}_finetuned\",\n",
    "        learning_rate=2e-5,\n",
    "        per_device_train_batch_size=8,\n",
    "        per_device_eval_batch_size=8,\n",
    "        num_train_epochs=5,  # Early stopping will control this\n",
    "        weight_decay=0.01,\n",
    "        eval_strategy=\"epoch\",\n",
    "        save_strategy=\"epoch\",\n",
    "        logging_strategy=\"steps\",\n",
    "        logging_steps=50,\n",
    "        evaluation_strategy=\"epoch\",\n",
    "        load_best_model_at_end=True,\n",
    "        metric_for_best_model=\"eval_f1_macro\",\n",
    "        greater_is_better=True,\n",
    "        save_total_limit=3,  # Keep only the 3 most recent checkpoints\n",
    "        gradient_accumulation_steps=2,  # To simulate a larger batch size\n",
    "        lr_scheduler_type=\"cosine\",  # Learning rate scheduler\n",
    "        push_to_hub=False,\n",
    "        report_to=\"none\"  # Suppress unwanted logging to services like WandB if not using them\n",
    "    )\n",
    "\n",
    "    # Define the metric computation function for macro metrics\n",
    "    def compute_metrics(eval_pred):\n",
    "        logits, labels = eval_pred\n",
    "        predictions = np.argmax(logits, axis=-1)\n",
    "        precision, recall, f1, _ = precision_recall_fscore_support(labels, predictions, average=\"macro\")\n",
    "        acc = accuracy_score(labels, predictions)\n",
    "        return {\"accuracy\": acc, \"f1_macro\": f1, \"precision_macro\": precision, \"recall_macro\": recall}\n",
    "\n",
    "    # Instantiate the Trainer with early stopping\n",
    "    trainer = Trainer(\n",
    "        model=model,\n",
    "        args=training_args,\n",
    "        train_dataset=train_dataset,\n",
    "        eval_dataset=test_dataset,\n",
    "        compute_metrics=compute_metrics,\n",
    "        callbacks=[EarlyStoppingCallback(early_stopping_patience=2, early_stopping_threshold=0.01)]\n",
    "    )\n",
    "\n",
    "    # Train the model\n",
    "    train_output = trainer.train()\n",
    "\n",
    "    # Print training metrics\n",
    "    print(\"\\nTraining output summary:\")\n",
    "    print(train_output)\n",
    "\n",
    "    # Evaluate the model on the test set\n",
    "    eval_results = trainer.evaluate()\n",
    "    print(f\"Evaluation results for {model_id}: {eval_results}\")\n",
    "\n",
    "    # Additional debug: Check final train/eval loss\n",
    "    print(f\"Final train loss: {train_output.training_loss}\")\n",
    "    print(f\"Final evaluation loss: {eval_results['eval_loss']}\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# train_and_evaluate('distilbert-base-german-cased')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# train_and_evaluate('bert-base-german-cased')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# train_and_evaluate('xlm-roberta-base')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# train_and_evaluate('german-nlp-group/electra-base-german-uncased')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# train_and_evaluate('nreimers/mMiniLMv2-L6-H384-distilled-from-XLMR-Large')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Summary of the Variants\n",
    "\n",
    "DistilBERT and BERT models are standard choices for German text classification.\n",
    "\n",
    "XLM-Roberta offers a robust multilingual alternative.\n",
    "\n",
    "Electra models are optimized for efficiency, with small and base versions for different speed and performance trade-offs.\n",
    "\n",
    "MiniLM provides an even more efficient option, suitable for quick benchmarking."
   ]
  },
  {
   "attachments": {
    "image.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![image.png](attachment:image.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import AutoModelForSequenceClassification, Trainer\n",
    "from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay, roc_curve, auc\n",
    "import torch\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "def evaluate_checkpoint(checkpoint_path, model_id):\n",
    "    # Prepare the datasets (make sure the same train/test split and tokenization as before)\n",
    "    train_dataset, test_dataset = prepare_dataset(train_texts, test_texts, train_labels, test_labels, model_id)\n",
    "\n",
    "    # Load the model from the checkpoint\n",
    "    model = AutoModelForSequenceClassification.from_pretrained(checkpoint_path)\n",
    "    device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "    model.to(device)\n",
    "\n",
    "    # Instantiate the Trainer with the loaded model\n",
    "    trainer = Trainer(\n",
    "        model=model,\n",
    "        eval_dataset=test_dataset\n",
    "    )\n",
    "\n",
    "    # Evaluate the model and generate predictions\n",
    "    predictions_output = trainer.predict(test_dataset)\n",
    "    logits = predictions_output.predictions\n",
    "    predicted_labels = np.argmax(logits, axis=-1)\n",
    "    predicted_probs = torch.nn.functional.softmax(torch.tensor(logits), dim=1)[:, 1].numpy()  # Probability for the positive class\n",
    "\n",
    "    # Actual labels\n",
    "    true_labels = predictions_output.label_ids\n",
    "\n",
    "    # Generate and plot the confusion matrix\n",
    "    cm = confusion_matrix(true_labels, predicted_labels)\n",
    "    disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=[\"Not Relevant\", \"Relevant\"])\n",
    "    disp.plot(cmap=plt.cm.Blues)\n",
    "    plt.title(f'Confusion Matrix for {model_id} at checkpoint {checkpoint_path}')\n",
    "    plt.show()\n",
    "\n",
    "    # Calculate and plot the ROC curve\n",
    "    fpr, tpr, thresholds = roc_curve(true_labels, predicted_probs)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "\n",
    "    plt.figure()\n",
    "    plt.plot(fpr, tpr, color='blue', lw=2, label=f'ROC curve (area = {roc_auc:.2f})')\n",
    "    plt.plot([0, 1], [0, 1], color='grey', lw=2, linestyle='--')\n",
    "    plt.xlim([0.0, 1.0])\n",
    "    plt.ylim([0.0, 1.05])\n",
    "    plt.xlabel('False Positive Rate')\n",
    "    plt.ylabel('True Positive Rate')\n",
    "    plt.title(f'ROC Curve for {model_id} at checkpoint {checkpoint_path}')\n",
    "    plt.legend(loc=\"lower right\")\n",
    "    plt.show()\n",
    "\n",
    "# Example usage:\n",
    "# checkpoint_path = \"distilbert-base-german-cased_finetuned/checkpoint-570\"\n",
    "# evaluate_checkpoint(checkpoint_path, 'distilbert-base-german-cased')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Initial train dataset sample:\n",
      "{'text': 'Seht ihr, Apokalypse, egal, worüber man sich den Kopf zerbricht.', 'label': 0, '__index_level_0__': 3643}\n",
      "Initial test dataset sample:\n",
      "{'text': 'Die KI könnte etwa bei Entscheidungen über die Einstellung von Personal oder in der Verwaltung als mögliches Korrektiv oder zweite Meinung zum Einsatz kommen, ohne dass die Entscheidung an das System delegiert würde.', 'label': 1, '__index_level_0__': 1011}\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "07d905a4b7184ffe9e98a78dcd1d27e5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Map:   0%|          | 0/3047 [00:00<?, ? examples/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b1885581b05f4476bb93de83eef31232",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Map:   0%|          | 0/762 [00:00<?, ? examples/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Tokenized train dataset sample:\n",
      "{'text': 'Seht ihr, Apokalypse, egal, worüber man sich den Kopf zerbricht.', 'labels': 0, '__index_level_0__': 3643, 'input_ids': [102, 23088, 348, 818, 5035, 3932, 1193, 237, 818, 8213, 818, 5928, 278, 545, 251, 190, 3263, 5666, 28749, 566, 103, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], 'attention_mask': [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]}\n",
      "Tokenized test dataset sample:\n",
      "{'text': 'Die KI könnte etwa bei Entscheidungen über die Einstellung von Personal oder in der Verwaltung als mögliches Korrektiv oder zweite Meinung zum Einsatz kommen, ohne dass die Entscheidung an das System delegiert würde.', 'labels': 1, '__index_level_0__': 1011, 'input_ids': [102, 229, 157, 30922, 1882, 1096, 282, 7048, 304, 128, 9770, 195, 4960, 394, 153, 125, 4401, 276, 9090, 30886, 3406, 203, 1012, 394, 3246, 5334, 386, 2089, 1263, 818, 1236, 377, 128, 2265, 208, 199, 2015, 25961, 11334, 30885, 1618, 566, 103, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], 'attention_mask': [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]}\n",
      "\n",
      "Final train dataset sample after format:\n",
      "{'labels': tensor(0), 'input_ids': tensor([  102, 23088,   348,   818,  5035,  3932,  1193,   237,   818,  8213,\n",
      "          818,  5928,   278,   545,   251,   190,  3263,  5666, 28749,   566,\n",
      "          103,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0]), 'attention_mask': tensor([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0])}\n",
      "Final test dataset sample after format:\n",
      "{'labels': tensor(1), 'input_ids': tensor([  102,   229,   157, 30922,  1882,  1096,   282,  7048,   304,   128,\n",
      "         9770,   195,  4960,   394,   153,   125,  4401,   276,  9090, 30886,\n",
      "         3406,   203,  1012,   394,  3246,  5334,   386,  2089,  1263,   818,\n",
      "         1236,   377,   128,  2265,   208,   199,  2015, 25961, 11334, 30885,\n",
      "         1618,   566,   103,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0,     0,     0,     0,     0,     0,     0,     0,     0,\n",
      "            0,     0]), 'attention_mask': tensor([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
      "        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
      "        0, 0, 0, 0, 0, 0, 0, 0])}\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6577d402868e4d04ae094e2d97b0d95d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/96 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "checkpoint_path = \"distilbert-base-german-cased_finetuned/checkpoint-570\"\n",
    "evaluate_checkpoint(checkpoint_path, 'distilbert-base-german-cased')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Text: \"Dieses autonome Auto verwendet KI, um die Sicherheit auf den Straßen zu verbessern.\"\n",
    "# Text: \"Der Kaffee war heute Morgen besonders gut im neuen Café um die Ecke.\"\n",
    "# Text: \"Ein neues Projekt zur Entwicklung nachhaltiger Technologien wurde in München gestartet.\"\n",
    "# Text: \"Die Wettervorhersage für morgen ist Regen und niedrige Temperaturen.\"\n",
    "\n",
    "\n",
    "# Text: \"Forscher haben eine innovative Methode zur Verbesserung der KI-basierten Sprachverarbeitung entwickelt.\"\n",
    "# Text: \"Das Restaurant hat jetzt eine neue Mittagskarte mit veganen Optionen.\"\n",
    "# Text: \"In Berlin wurde eine Konferenz über maschinelles Lernen und dessen Anwendungen in der Gesundheitsbranche abgehalten.\"\n",
    "\n",
    "# Text: \"Die Fußballmannschaft hat das Spiel gestern Abend mit 2:1 gewonnen.\"\n",
    "# Text: \"Mit dem neuen Algorithmus können Unternehmen ihre Daten effizienter analysieren.\"\n",
    "# Text: \"Der Film, den ich am Wochenende gesehen habe, war sehr spannend.\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from transformers import AutoTokenizer, AutoModelForSequenceClassification\n",
    "import torch.nn.functional as F\n",
    "import numpy as np\n",
    "\n",
    "def load_model_and_tokenizer(checkpoint_path, model_id):\n",
    "    # Load the tokenizer specific to the model\n",
    "    tokenizer = AutoTokenizer.from_pretrained(model_id)\n",
    "\n",
    "    # Load the model from the checkpoint\n",
    "    model = AutoModelForSequenceClassification.from_pretrained(checkpoint_path)\n",
    "    device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "    model.to(device)\n",
    "    model.eval()\n",
    "\n",
    "    return model, tokenizer, device\n",
    "\n",
    "def predict_text(model, tokenizer, device, text):\n",
    "    # Tokenize the input text\n",
    "    inputs = tokenizer(text, return_tensors=\"pt\", padding=True, truncation=True, max_length=512)\n",
    "    inputs = {k: v.to(device) for k, v in inputs.items()}\n",
    "\n",
    "    # Get the model output\n",
    "    with torch.no_grad():\n",
    "        outputs = model(**inputs)\n",
    "        logits = outputs.logits\n",
    "\n",
    "    # Apply softmax to get probabilities\n",
    "    probabilities = F.softmax(logits, dim=-1).cpu().numpy()[0]\n",
    "    return probabilities\n",
    "\n",
    "def main():\n",
    "    checkpoint_path = \"distilbert-base-german-cased_finetuned/checkpoint-570\"\n",
    "    model_id = 'distilbert-base-german-cased'\n",
    "    \n",
    "    # Load model and tokenizer\n",
    "    model, tokenizer, device = load_model_and_tokenizer(checkpoint_path, model_id)\n",
    "    \n",
    "    print(\"Model loaded. Type a sentence to get a prediction or type 'exit' to quit.\")\n",
    "\n",
    "    while True:\n",
    "        # Get user input\n",
    "        text = input(\"\\nEnter a text: \")\n",
    "\n",
    "        if text.lower() == 'exit':\n",
    "            print(\"Exiting the app. Goodbye!\")\n",
    "            break\n",
    "\n",
    "        # Get prediction\n",
    "        probabilities = predict_text(model, tokenizer, device, text)\n",
    "\n",
    "        # Display the result\n",
    "        print(f\"\\nPrediction probabilities:\")\n",
    "        print(f\"Not Relevant: {probabilities[0]:.4f}\")\n",
    "        print(f\"Relevant: {probabilities[1]:.4f}\")\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    main()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "wsk",
   "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": "3.12.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}