modern-bert-seq-class-values-no-context_plus

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

• Loss: 0.4721
• Subset Accuracy: 0.2877
• F1 Macro: 0.3277
• F1 Micro: 0.3949
• Precision Macro: 0.4016
• Recall Macro: 0.2853
• Roc Auc: 0.8036

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 2025
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 25
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Subset Accuracy	F1 Macro	F1 Micro	Precision Macro	Recall Macro	Roc Auc
0.4892	1.0	767	0.1669	0.1642	0.1686	0.2704	0.4781	0.1215	0.8258
0.3063	2.0	1534	0.1582	0.2215	0.2160	0.3381	0.6175	0.1601	0.8524
0.2401	3.0	2301	0.1640	0.2410	0.2764	0.3530	0.4702	0.2059	0.8437
0.1323	4.0	3068	0.2049	0.2792	0.3119	0.3880	0.4098	0.2720	0.8240
0.0739	5.0	3835	0.2595	0.2836	0.3159	0.3939	0.4256	0.2737	0.8177
0.039	6.0	4602	0.2987	0.2799	0.3054	0.3942	0.4040	0.2615	0.8118
0.0249	7.0	5369	0.3283	0.2688	0.3133	0.3878	0.4067	0.2674	0.8116
0.0145	8.0	6136	0.3720	0.2694	0.3194	0.3941	0.3892	0.2965	0.8063
0.0109	9.0	6903	0.3840	0.2803	0.3197	0.4014	0.4034	0.2790	0.8075
0.0073	10.0	7670	0.4166	0.2746	0.3278	0.3892	0.3914	0.2942	0.8083
0.007	11.0	8437	0.4251	0.2773	0.3231	0.3861	0.4175	0.2749	0.8053
0.0046	12.0	9204	0.4406	0.2848	0.3220	0.3975	0.4023	0.2820	0.8065
0.0046	13.0	9971	0.4518	0.2794	0.3154	0.3872	0.4223	0.2615	0.8019
0.0044	14.0	10738	0.4822	0.2784	0.3119	0.3859	0.4071	0.2632	0.7930
0.0042	15.0	11505	0.4721	0.2877	0.3277	0.3949	0.4016	0.2853	0.8036

Framework versions

• Transformers 4.53.2
• Pytorch 2.6.0+cu124
• Datasets 2.14.4
• Tokenizers 0.21.2