r-f/ModernBERT-large-zeroshot-v1

Model Description

This model is a fine-tuned ModernBERT-large for Natural Language Inference. It was trained on the MoritzLaurer/synthetic_zeroshot_mixtral_v0.1 and is designed to carry out zero-shot classification.

Model Overview

• Model Type: ModernBERT-large (BERT variant)
• Task: Zero-shot Classification
• Languages: English
• Dataset: MoritzLaurer/synthetic_zeroshot_mixtral_v0.1
• Fine-Tuning: Fine-tuned for Zero-shot Classification

Performance Metrics

To be added.

• Training Loss: Measures the model's fit to the training data.
• Validation Loss: Measures the model's generalization to unseen data.
• Accuracy: The percentage of correct predictions over all examples.
• F1 Score: A balanced metric between precision and recall.

Installation and Example Usage

pip install transformers torch datasets

classifier = pipeline("zero-shot-classification", "r-f/ModernBERT-large-zeroshot-v1")
sequence_to_classify = "I want to be an actor."
candidate_labels = ["space", "economy", "entertainment"]
output = classifier(sequence_to_classify, candidate_labels, multi_label=False)
print(output)
>>{'sequence': 'I want to be an actor.', 'labels': ['entertainment', 'space', 'economy'], 'scores': [0.9614731073379517, 0.028852475807070732, 0.009674412198364735]}

Model Card

• Model Name: ModernBERT-large-zeroshot-v1
• Hugging Face Repo: r-f/ModernBERT-large-zeroshot-v1
• License: MIT (or another applicable license)
• Date: 23-12-2024

Training Details

• Model: ModernBERT (Large variant)
• Framework: PyTorch
• Batch Size: 32
• Learning Rate: 2e-5
• Optimizer: AdamW
• Hardware: RTX 4090

Acknowledgments

• The model was trained on the MoritzLaurer/synthetic_zeroshot_mixtral_v0.1. And the training script was adapted from MoritzLaurer/zeroshot-classifier
• Special thanks to the Hugging Face community and all contributors to the transformers library.

License

This model is licensed under the MIT License. See the LICENSE file for more details.