biotech-sentiment-modernbert-large

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

• Loss: 4.3531
• F1 Weighted: 0.5360
• F1 Macro: 0.4102
• F1 Micro: 0.5577
• Accuracy: 0.5577
• F1 Class 0: 0.2881
• F1 Class 1: 0.7143
• F1 Class 2: 0.2281

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: 16
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 250
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Weighted	F1 Macro	F1 Micro	Accuracy	F1 Class 0	F1 Class 1	F1 Class 2
2.1818	1.0	127	1.0826	0.3588	0.3214	0.3504	0.3504	0.3487	0.4120	0.2034
2.2104	2.0	254	1.1365	0.2625	0.2229	0.2991	0.2991	0.3404	0.3185	0.0100
2.0356	3.0	381	1.2092	0.3658	0.2829	0.3722	0.3722	0.0293	0.4832	0.3363
1.6188	4.0	508	1.1989	0.3326	0.3157	0.3198	0.3198	0.2950	0.3567	0.2952
1.1139	5.0	635	1.5419	0.5282	0.4009	0.5479	0.5479	0.2277	0.7086	0.2663
0.7829	6.0	762	2.0069	0.5263	0.4022	0.5410	0.5410	0.3171	0.7022	0.1871
0.5745	7.0	889	2.6626	0.5037	0.3840	0.5035	0.5035	0.1843	0.6733	0.2943
0.4521	8.0	1016	2.7618	0.5304	0.4041	0.5508	0.5508	0.2569	0.7094	0.2460
0.3251	9.0	1143	2.9332	0.5332	0.3932	0.5706	0.5706	0.2360	0.7317	0.2118
0.2154	10.0	1270	2.9890	0.5285	0.4127	0.5350	0.5350	0.3134	0.6927	0.2321
0.1807	11.0	1397	3.2419	0.5279	0.4132	0.5321	0.5321	0.2995	0.6906	0.2493
0.0934	12.0	1524	3.5012	0.5121	0.4031	0.5104	0.5104	0.3230	0.6667	0.2196
0.0738	13.0	1651	4.1345	0.5332	0.4017	0.5607	0.5607	0.2943	0.7197	0.1911
0.0107	14.0	1778	4.0537	0.5337	0.4117	0.5499	0.5499	0.3128	0.7066	0.2156
0.0141	15.0	1905	4.0290	0.5199	0.3972	0.5301	0.5301	0.2948	0.6939	0.2029
0.0067	16.0	2032	4.3166	0.5352	0.4112	0.5558	0.5558	0.2992	0.7109	0.2235
0.0024	17.0	2159	4.3491	0.5359	0.4101	0.5577	0.5577	0.2865	0.7143	0.2294
0.0101	18.0	2286	4.3545	0.5352	0.4096	0.5568	0.5568	0.2881	0.7133	0.2274
0.0025	19.0	2413	4.3476	0.5354	0.4095	0.5568	0.5568	0.2873	0.7139	0.2274
0.0091	20.0	2540	4.3531	0.5360	0.4102	0.5577	0.5577	0.2881	0.7143	0.2281

Framework versions

• Transformers 5.3.0
• Pytorch 2.10.0+cu128
• Datasets 4.6.1
• Tokenizers 0.22.2