modern-bert-seq-class-values-no-context_cru

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: 0.3111
• Subset Accuracy: 0.2931
• F1 Macro: 0.3240
• F1 Micro: 0.4090
• Precision Macro: 0.4089
• Recall Macro: 0.2766
• Roc Auc: 0.8240

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 2025
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• 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: 20
• 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
1.1484	0.5737	767	0.1752	0.1041	0.0849	0.1841	0.2599	0.0598	0.7853
0.6467	1.1473	1534	0.1628	0.2092	0.2078	0.3286	0.5448	0.1682	0.8450
0.604	1.7210	2301	0.1583	0.2337	0.2625	0.3514	0.5243	0.2046	0.8559
0.4872	2.2947	3068	0.1659	0.3022	0.3181	0.4250	0.4744	0.2649	0.8523
0.4325	2.8684	3835	0.1642	0.2970	0.3079	0.4165	0.4808	0.2495	0.8549
0.2084	3.4420	4602	0.2129	0.3120	0.3389	0.4253	0.4198	0.3042	0.8402
0.2093	4.0157	5369	0.2313	0.3136	0.3324	0.4233	0.4264	0.2844	0.8390
0.0995	4.5894	6136	0.2614	0.2967	0.3277	0.4182	0.3957	0.2941	0.8276
0.1017	5.1631	6903	0.2819	0.2820	0.3189	0.3997	0.4125	0.2734	0.8282
0.0569	5.7367	7670	0.2997	0.2930	0.3285	0.4217	0.4055	0.2898	0.8236
0.0628	6.3104	8437	0.3111	0.2931	0.3240	0.4090	0.4089	0.2766	0.8240

Framework versions

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