modernbert-base-multi-head-values-context-roc_auc

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: 0.1835
• Subset Accuracy: 0.3032
• F1 Macro: 0.3352
• F1 Micro: 0.4194
• Precision Macro: 0.4748
• Recall Macro: 0.2755
• Roc Auc: 0.8264

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-06
• train_batch_size: 2
• eval_batch_size: 2
• seed: 2025
• gradient_accumulation_steps: 8
• 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: 0.01
• num_epochs: 33
• 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
2.6266	0.5002	767	0.2033	0.0069	0.0054	0.0125	0.0347	0.0029	0.6535
1.5099	1.0	1534	0.1838	0.0815	0.0668	0.1394	0.2799	0.0412	0.7493
1.4376	1.5002	2301	0.1765	0.1326	0.1239	0.2181	0.3987	0.0827	0.7851
1.3726	2.0	3068	0.1680	0.1946	0.1786	0.2948	0.4573	0.1265	0.8058
1.2975	2.5002	3835	0.1653	0.2098	0.1955	0.3157	0.4924	0.1455	0.8188
1.2668	3.0	4602	0.1598	0.2547	0.2460	0.3660	0.5225	0.1833	0.8312
1.2172	3.5002	5369	0.1579	0.2364	0.2546	0.3487	0.5852	0.1808	0.8361
1.1801	4.0	6136	0.1554	0.2426	0.2481	0.3576	0.6564	0.1772	0.8417
1.1193	4.5002	6903	0.1562	0.2773	0.2924	0.3950	0.5843	0.2169	0.8420
1.1036	5.0	7670	0.1555	0.2656	0.2821	0.3803	0.5656	0.2047	0.8451
1.0424	5.5002	8437	0.1578	0.3004	0.3118	0.4169	0.5477	0.2396	0.8442
1.0105	6.0	9204	0.1586	0.2747	0.3026	0.3951	0.5986	0.2289	0.8436
0.9595	6.5002	9971	0.1628	0.3067	0.3102	0.4204	0.5569	0.2437	0.8399
0.9071	7.0	10738	0.1627	0.2924	0.3178	0.4078	0.5495	0.2397	0.8394
0.8032	7.5002	11505	0.1700	0.3056	0.3298	0.4215	0.5176	0.2636	0.8366
0.8074	8.0	12272	0.1688	0.2973	0.3267	0.4141	0.5121	0.2569	0.8347
0.6872	8.5002	13039	0.1763	0.3083	0.3363	0.4268	0.4972	0.2725	0.8331
0.6872	9.0	13806	0.1761	0.3008	0.3329	0.4197	0.4891	0.2651	0.8327
0.5634	9.5002	14573	0.1835	0.3032	0.3352	0.4194	0.4748	0.2755	0.8264

Framework versions

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