CaXLMRoBERTa-large_checkpoint-15596

Model Card for Model ID

We fine-tuned our base model for 71 epochs on the Ca dataset, epoch 61 showed the best macro average f1 score on the evaluation dataset.

Metrics

eval_AVGf1 0.8073040334161414

eval_DIAGNOSIS.f1 0.8044417026526834

eval_DIAGNOSIS.precision 0.7774244833068362

eval_DIAGNOSIS.recall 0.8334043459735833

eval_DIAGNOSTIC.f1 0.8154647655607348

eval_DIAGNOSTIC.precision 0.7876059322033898

eval_DIAGNOSTIC.recall 0.8453666856168277

eval_DRUG.f1 0.9283865401207938

eval_DRUG.precision 0.911864406779661

eval_DRUG.recall 0.945518453427065

eval_MEDICAL_FINDING.f1 0.7855789872458644

eval_MEDICAL_FINDING.precision 0.7687839841819081

eval_MEDICAL_FINDING.recall 0.8031241931319391

eval_THERAPY.f1 0.7026481715006304

eval_THERAPY.precision 0.6716489874638379

eval_THERAPY.recall 0.7366472765732417

eval_accuracy 0.9359328085693419

eval_f1 0.7922039763638145

eval_loss 0.6178462505340576

eval_precision 0.7703492063492063

eval_recall 0.8153349909280291

eval_runtime 107.4969

eval_samples_per_second 76.114

eval_steps_per_second 9.517

test_AVGf1 0.7654950023468019

test_DIAGNOSIS.f1 0.7317784256559767

test_DIAGNOSIS.precision 0.7442550037064493

test_DIAGNOSIS.recall 0.7197132616487455

test_DIAGNOSTIC.f1 0.7815242494226328

test_DIAGNOSTIC.precision 0.7779310344827586

test_DIAGNOSTIC.recall 0.7851508120649652

test_DRUG.f1 0.9199594731509625

test_DRUG.precision 0.9013898080741231

test_DRUG.recall 0.9393103448275862

test_MEDICAL_FINDING.f1 0.7348673770120154

test_MEDICAL_FINDING.precision 0.6987497305453761

test_MEDICAL_FINDING.recall 0.7749223045661009

test_THERAPY.f1 0.6593454864924225

test_THERAPY.precision 0.6414529914529915

test_THERAPY.recall 0.6782647989154993

test_accuracy 0.9244002381348251

test_f1 0.7459972552607502

test_loss 0.7649919986724854

test_precision 0.72469725586046

test_recall 0.7685872510899022

test_runtime 124.1668

test_samples_per_second 76.421

test_steps_per_second 9.56