German Zeroshot

🚀 German Zero-Shot Classification Model

This German zero-shot classification model is a fine - tuned version of deepset/gbert-large on the facebook/xnli German dataset. It provides an efficient way to classify text without the need for task - specific training data.

🚀 Quick Start

This German zero - shot classification model is a fine - tuned version of deepset/gbert-large on the Facebook/XNLI German dataset. It achieves the following results on the evaluation set:

• Loss: 0.4592
• Accuracy: 0.8486

✨ Features

• Zero - Shot Classification: Capable of classifying text into predefined labels without task - specific training.
• German Language Support: Specifically fine - tuned for German text, making it suitable for German - speaking applications.

📦 Installation

The model can be used with the transformers library. You can install it using the following command:

pip install transformers

💻 Usage Examples

Basic Usage

# Use a pipeline as a high-level helper
import torch
from transformers import pipeline

pipe = pipeline(
    "zero-shot-classification",
    model="kaixkhazaki/german-zeroshot",
    tokenizer="kaixkhazaki/german-zeroshot",
    device=0 if torch.cuda.is_available() else -1  # Use GPU if available
)

#Enter your text and possible candidates of classification
sequence = "Können Sie mir die Schritte zur Konfiguration eines VPN auf einem Linux-Server erklären?"
candidate_labels = [
    "Technische Dokumentation", 
    "IT-Support", 
    "Netzwerkadministration", 
    "Linux-Konfiguration", 
    "VPN-Setup"
]
print(pipe(sequence,candidate_labels))

Advanced Usage

# Another example
sequence = "Wie lautet die Garantiezeit für dieses Produkt?"
candidate_labels = [
    "Garantiebedingungen", 
    "Kundendienst", 
    "Produktdetails", 
    "Reklamation", 
    "Kaufberatung"
]
print(pipe(sequence,candidate_labels))

📚 Documentation

Training and Evaluation Data

The model is fine - tuned on the Facebook/XNLI German dataset. However, more detailed information about the dataset is yet to be provided.

Training Procedure

Training Hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e - 05
• train_batch_size: 64
• eval_batch_size: 32
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon = 1e - 08 and optimizer_args = No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 500
• num_epochs: 3

Training Results

Framework versions

• Transformers 4.48.0.dev0
• Pytorch 2.4.1+cu121
• Datasets 3.1.0
• Tokenizers 0.21.0

📄 License

This model is released under the MIT license.