BioMedRoBERTa-finetuned-ner-pablo-classifier-then-full

This model is a fine-tuned version of pabRomero/BioMedRoBERTa-finetuned-ner-pablo-just-classifier on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0824
• Precision: 0.7761
• Recall: 0.7831
• F1: 0.7796
• Accuracy: 0.9747

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: 512
• eval_batch_size: 512
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 2048
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine_with_restarts
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	0.9697	16	0.1075	0.7084	0.7084	0.7084	0.9691
No log	2.0	33	0.0972	0.7475	0.7397	0.7436	0.9712
No log	2.9697	49	0.0922	0.7402	0.7483	0.7442	0.9725
No log	4.0	66	0.0880	0.7618	0.7503	0.7560	0.9734
No log	4.9697	82	0.0868	0.7612	0.7536	0.7573	0.9736
No log	6.0	99	0.0865	0.7601	0.7572	0.7586	0.9737
No log	6.9697	115	0.0863	0.7607	0.7588	0.7598	0.9737
No log	8.0	132	0.0875	0.7513	0.7716	0.7613	0.9737
No log	8.9697	148	0.0823	0.7706	0.7687	0.7696	0.9745
No log	10.0	165	0.0827	0.7625	0.7752	0.7688	0.9738
No log	10.9697	181	0.0824	0.7690	0.7739	0.7715	0.9746
No log	12.0	198	0.0818	0.7739	0.7739	0.7739	0.9748
No log	12.9697	214	0.0820	0.7718	0.7747	0.7732	0.9747
No log	14.0	231	0.0818	0.7735	0.7773	0.7754	0.9749
No log	14.9697	247	0.0820	0.7837	0.7757	0.7797	0.9754
No log	16.0	264	0.0831	0.7734	0.7842	0.7788	0.9749
No log	16.9697	280	0.0826	0.7683	0.7883	0.7782	0.9745
No log	18.0	297	0.0826	0.7747	0.7835	0.7791	0.9747
No log	18.9697	313	0.0824	0.7760	0.7830	0.7795	0.9747
No log	19.3939	320	0.0824	0.7761	0.7831	0.7796	0.9747

Framework versions

• Transformers 4.44.1
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1