disfluency_model

This model is a fine-tuned version of answerdotai/ModernBERT-large on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0048
• Accuracy: 0.9956
• Precision: 0.9847
• Recall: 0.9927
• F1: 0.9887

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_8BIT 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: 10
• mixed_precision_training: Native AMP
• label_smoothing_factor: 0.05

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.0044	1.2745	5000	0.0066	0.9865	0.9513	0.9786	0.9648
0.0058	2.5491	10000	0.0052	0.9932	0.9747	0.9873	0.9810
0.0036	3.8236	15000	0.0047	0.9888	0.9623	0.9846	0.9733
0.0005	5.0981	20000	0.0047	0.9943	0.9807	0.9904	0.9855
0.0011	6.3727	25000	0.0047	0.9946	0.9808	0.9903	0.9855
0.0016	7.6472	30000	0.0054	0.9951	0.9836	0.9909	0.9872
0.0001	8.9217	35000	0.0061	0.9952	0.9838	0.9905	0.9871

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.4.1
• Tokenizers 0.21.1