Model Card for Model ID

We fine-tuned our base model for 71 epochs on the Ca dataset, epoch 68 showed the best macro average f1 score on the evaluation dataset.

Metrics

eval_AVGf1 0.8032336746529752

eval_DIAGNOSIS.f1 0.7955801104972375

eval_DIAGNOSIS.precision 0.7656557699881843

eval_DIAGNOSIS.recall 0.82793867120954

eval_DIAGNOSTIC.f1 0.8097188097188096

eval_DIAGNOSTIC.precision 0.7797055730809674

eval_DIAGNOSTIC.recall 0.8421351504826803

eval_DRUG.f1 0.9214929214929215

eval_DRUG.precision 0.9002514668901928

eval_DRUG.recall 0.9437609841827768

eval_MEDICAL_FINDING.f1 0.7812833218340337

eval_MEDICAL_FINDING.precision 0.7604395604395604

eval_MEDICAL_FINDING.recall 0.8033019476331743

eval_THERAPY.f1 0.7080932097218742

eval_THERAPY.precision 0.6731777036684136

eval_THERAPY.recall 0.7468287526427061

eval_accuracy 0.9415681083480303

eval_f1 0.788057764075937

eval_loss 0.46635299921035767

eval_precision 0.7625447465929787

eval_recall 0.8153370937416062

eval_runtime 36.5944

eval_samples_per_second 223.586

eval_steps_per_second 27.955

test_AVGf1 0.765773820622575

test_DIAGNOSIS.f1 0.7267739575713241

test_DIAGNOSIS.precision 0.742803738317757

test_DIAGNOSIS.recall 0.711421410669531

test_DIAGNOSTIC.f1 0.7813144034806503

test_DIAGNOSTIC.precision 0.77124773960217

test_DIAGNOSTIC.recall 0.7916473317865429

test_DRUG.f1 0.9209993247805537

test_DRUG.precision 0.9021164021164021

test_DRUG.recall 0.9406896551724138

test_MEDICAL_FINDING.f1 0.7354366197183099

test_MEDICAL_FINDING.precision 0.6959164089988271

test_MEDICAL_FINDING.recall 0.7797156851033329

test_THERAPY.f1 0.6643447975620373

test_THERAPY.precision 0.6411764705882353

test_THERAPY.recall 0.6892502258355917

test_accuracy 0.9330358352068041

test_f1 0.7461369909791981

test_loss 0.5957663655281067

test_precision 0.7219958145170173

test_recall 0.7719484190072425

test_runtime 42.5823

test_samples_per_second 222.839

test_steps_per_second 27.875

Downloads last month
21
Safetensors
Model size
108M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for MSey/CaMedBERT-512_fl32_checkpoint-17386

Finetuned
(5)
this model

Collection including MSey/CaMedBERT-512_fl32_checkpoint-17386