modernbert-parap-10m-from-110m-labled

This model is a fine-tuned version of answerdotai/ModernBERT-large on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2753
• Accuracy: 0.8847
• F1: 0.8847
• Precision: 0.8848
• Recall: 0.8847
• F1 Class 0: 0.8849
• Precision Class 0: 0.8841
• Recall Class 0: 0.8857
• F1 Class 1: 0.8846
• Precision Class 1: 0.8854
• Recall Class 1: 0.8838

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 300
• eval_batch_size: 300
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• total_train_batch_size: 2400
• total_eval_batch_size: 2400
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1
• label_smoothing_factor: 0.01

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall	F1 Class 0	Precision Class 0	Recall Class 0	F1 Class 1	Precision Class 1	Recall Class 1
2.2266	1.0	7040	0.2753	0.8847	0.8847	0.8848	0.8847	0.8849	0.8841	0.8857	0.8846	0.8854	0.8838

Framework versions

• Transformers 4.55.2
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4