NeRUBioS_RoBERTa_Training_Testing

This model is a fine-tuned version of PlanTL-GOB-ES/roberta-base-biomedical-clinical-es on an adaptation of the Nubes dataset called NeRUBioS. Training and Testing Datasets have 13832 and 2765 samples, respectively. This is a result of the PhD dissertation of Antonio Tamayo. It achieves the following results on the evaluation set:

• Loss: 0.3540
• Negref Precision: 0.5522
• Negref Recall: 0.6138
• Negref F1: 0.5814
• Neg Precision: 0.9530
• Neg Recall: 0.9684
• Neg F1: 0.9606
• Nsco Precision: 0.8812
• Nsco Recall: 0.9092
• Nsco F1: 0.8950
• Unc Precision: 0.8208
• Unc Recall: 0.8923
• Unc F1: 0.8550
• Usco Precision: 0.6786
• Usco Recall: 0.7815
• Usco F1: 0.7264
• Precision: 0.8223
• Recall: 0.8680
• F1: 0.8446
• Accuracy: 0.9497

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 12

Training results

Training Loss	Epoch	Step	Validation Loss	Negref Precision	Negref Recall	Negref F1	Neg Precision	Neg Recall	Neg F1	Nsco Precision	Nsco Recall	Nsco F1	Unc Precision	Unc Recall	Unc F1	Usco Precision	Usco Recall	Usco F1	Precision	Recall	F1	Accuracy
0.1945	1.0	1729	0.2026	0.4356	0.4919	0.4621	0.9246	0.9558	0.9399	0.8264	0.9039	0.8634	0.7192	0.8077	0.7609	0.6125	0.7069	0.6563	0.7610	0.8276	0.7929	0.9383
0.1173	2.0	3458	0.2070	0.4961	0.5595	0.5259	0.9314	0.9635	0.9472	0.8552	0.9070	0.8803	0.8221	0.8769	0.8486	0.6472	0.7404	0.6906	0.7953	0.8516	0.8225	0.9423
0.0938	3.0	5187	0.2145	0.5006	0.6182	0.5532	0.9348	0.9670	0.9506	0.8679	0.9047	0.8859	0.7991	0.8667	0.8315	0.6316	0.7404	0.6817	0.7919	0.8607	0.8249	0.9435
0.0543	4.0	6916	0.2151	0.5267	0.6226	0.5707	0.9425	0.9677	0.9550	0.8588	0.9017	0.8797	0.8076	0.8718	0.8385	0.6733	0.7789	0.7223	0.8036	0.8647	0.8330	0.9460
0.0403	5.0	8645	0.2669	0.5627	0.5536	0.5581	0.9470	0.9670	0.9569	0.8697	0.9085	0.8886	0.8153	0.8487	0.8317	0.7009	0.7712	0.7344	0.8265	0.8526	0.8393	0.9472
0.0325	6.0	10374	0.2608	0.5207	0.6094	0.5616	0.9503	0.9677	0.9589	0.8751	0.9062	0.8904	0.7977	0.8795	0.8366	0.6771	0.7815	0.7255	0.8093	0.8650	0.8362	0.9476
0.0205	7.0	12103	0.3006	0.5646	0.6285	0.5949	0.9419	0.9684	0.9550	0.8671	0.9130	0.8895	0.7958	0.8795	0.8356	0.6594	0.7815	0.7153	0.8125	0.8704	0.8404	0.9484
0.0146	8.0	13832	0.3124	0.5653	0.5844	0.5747	0.9569	0.9677	0.9623	0.8896	0.9085	0.8990	0.8085	0.8769	0.8413	0.6966	0.7969	0.7434	0.8320	0.8628	0.8471	0.9496
0.0095	9.0	15561	0.3160	0.5459	0.6285	0.5843	0.9544	0.9705	0.9624	0.8770	0.9115	0.8939	0.8255	0.8974	0.8600	0.6868	0.7892	0.7344	0.8202	0.8730	0.8458	0.9511
0.0069	10.0	17290	0.3415	0.5461	0.6094	0.5760	0.9536	0.9677	0.9606	0.8810	0.9070	0.8938	0.8182	0.9	0.8571	0.6785	0.7866	0.7286	0.8207	0.8676	0.8435	0.9505
0.0047	11.0	19019	0.3483	0.5481	0.6197	0.5817	0.9517	0.9684	0.9600	0.8776	0.9107	0.8938	0.8160	0.8872	0.8501	0.6909	0.7815	0.7334	0.8204	0.8690	0.8440	0.9490
0.0018	12.0	20748	0.3540	0.5522	0.6138	0.5814	0.9530	0.9684	0.9606	0.8812	0.9092	0.8950	0.8208	0.8923	0.8550	0.6786	0.7815	0.7264	0.8223	0.8680	0.8446	0.9497

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.3