CeLLaTe-tapt-pubmedbert-tokenizer-adapted-baseline-combinedData

This model is a fine-tuned version of microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext on the Mardiyyah/Combined_TAPT_dataset dataset. It achieves the following results on the evaluation set:

Model description

More information needed

More information needed

More information needed

The following hyperparameters were used during training:

learning_rate: 3e-05
train_batch_size: 32
eval_batch_size: 32
seed: 3407
gradient_accumulation_steps: 2
total_train_batch_size: 64
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-06 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.06
num_epochs: 100
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Accuracy	Perplexity
1.3987	1.0	28	1.2896	0.7525	3.6312
1.3501	2.0	56	1.2327	0.7561	3.4304
1.3044	3.0	84	1.2182	0.7552	3.3810
1.2667	4.0	112	1.1852	0.7573	3.2712
1.2469	5.0	140	1.1747	0.7596	3.2372
1.2284	6.0	168	1.1602	0.7603	3.1906
1.1974	7.0	196	1.1517	0.7599	3.1635
1.1961	8.0	224	1.1234	0.7636	3.0752
1.1581	9.0	252	1.1104	0.7694	3.0356
1.1443	10.0	280	1.1184	0.7687	3.0600
1.1272	11.0	308	1.1149	0.7632	3.0494
1.1141	12.0	336	1.1268	0.7655	3.0858
1.114	13.0	364	1.1188	0.7634	3.0612
1.0733	14.0	392	1.0972	0.7661	2.9959
1.0756	15.0	420	1.1112	0.7665	3.0381
1.0767	16.0	448	1.1173	0.7616	3.0566
1.0561	17.0	476	1.0940	0.7678	2.9861
1.0528	18.0	504	1.0975	0.7640	2.9968
1.0363	19.0	532	1.0983	0.7658	2.9991
1.0247	20.0	560	1.0869	0.7684	2.9652
1.0185	21.0	588	1.1124	0.7635	3.0416
1.008	22.0	616	1.1290	0.7593	3.0926
0.9954	23.0	644	1.1142	0.7607	3.0472
1.0192	24.0	672	1.1127	0.7619	3.0425
1.0038	25.0	700	1.1115	0.7642	3.0389