Bart Large Paraphrase Generator En De V2

🚀 bart_large_paraphrase_generator_en_de_v2

This is a model trained from scratch on an unknown dataset, which can potentially be used for tasks like paraphrase generation.

🚀 Quick Start

There is no specific quick - start guide provided in the original document.

📚 Documentation

Model description

This model was trained from scratch on an unknown dataset. More information is needed to fully understand its characteristics.

Intended uses & limitations

More information regarding the intended uses and limitations of this model is required.

Evaluation results

The following evaluation results are provided:

{
    "eval_loss": 0.9200083613395691,
    "eval_score": 49.97448884411352,
    "eval_counts": [100712, 72963, 57055, 41578],
    "eval_totals": [133837, 130839, 127841, 124843],
    "eval_precisions": [75.24974409169363, 55.76548276889918, 44.6296571522438, 33.30423011302196],
    "eval_bp": 1.0,
    "eval_sys_len": 133837,
    "eval_ref_len": 130883,
    "eval_runtime": 138.6871,
    "eval_samples_per_second": 21.617,
    "eval_steps_per_second": 0.678
}

However, more information is needed to interpret these results comprehensively.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e - 05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi - GPU
• num_devices: 4
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• total_eval_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
• lr_scheduler_type: linear
• num_epochs: 3.0

Framework versions

• Transformers 4.16.2
• Pytorch 1.11.0a0 + bfe5ad2
• Datasets 1.18.3
• Tokenizers 0.11.0