ModernBERT-large_v3

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: 0.8266
• Accuracy: 0.9109
• Precision Macro: 0.7681
• Recall Macro: 0.7438
• F1 Macro: 0.7542
• F1 Weighted: 0.9084

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision Macro	Recall Macro	F1 Macro	F1 Weighted
0.9719	1.0	179	0.4542	0.8484	0.7048	0.5982	0.5951	0.8301
0.6216	2.0	358	0.3472	0.8819	0.8897	0.6468	0.6648	0.8674
0.5377	3.0	537	0.3063	0.8926	0.7740	0.7326	0.7477	0.8893
0.4253	4.0	716	0.2703	0.9109	0.8330	0.7357	0.7651	0.9053
0.3699	5.0	895	0.2795	0.9090	0.7756	0.7968	0.7850	0.9107
0.2003	6.0	1074	0.3297	0.9128	0.8225	0.7620	0.7848	0.9094
0.1596	7.0	1253	0.3799	0.9097	0.7673	0.7805	0.7734	0.9109
0.0876	8.0	1432	0.5013	0.9236	0.8343	0.7899	0.8084	0.9214
0.0598	9.0	1611	0.5279	0.9185	0.8126	0.7621	0.7815	0.9152
0.054	10.0	1790	0.5909	0.9109	0.7998	0.7728	0.7847	0.9092
0.0419	11.0	1969	0.7661	0.9141	0.7877	0.7427	0.7594	0.9102
0.0108	12.0	2148	0.9184	0.9185	0.8260	0.7337	0.7601	0.9122
0.0177	13.0	2327	0.8254	0.9128	0.7820	0.7494	0.7628	0.9099
0.0013	14.0	2506	0.8059	0.9103	0.7741	0.7391	0.7531	0.9069
0.0019	15.0	2685	0.8174	0.9078	0.7620	0.7502	0.7556	0.9065
0.0028	16.0	2864	0.8202	0.9109	0.7704	0.7438	0.7550	0.9082
0.0	17.0	3043	0.8126	0.9103	0.7678	0.7433	0.7537	0.9078
0.0008	18.0	3222	0.8319	0.9109	0.7734	0.7482	0.7589	0.9085
0.0	19.0	3401	0.8245	0.9116	0.7686	0.7443	0.7546	0.9090
0.0001	20.0	3580	0.8266	0.9109	0.7681	0.7438	0.7542	0.9084

Framework versions

• Transformers 4.55.0
• Pytorch 2.7.0+cu126
• Datasets 4.0.0
• Tokenizers 0.21.4