CeLLaTe-tapt-pubmedbert-tokenizer-adapted-wwmask

This model is a fine-tuned version of microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext on the Mardiyyah/TAPT_CeLLaTe2.0 dataset. It achieves the following results on the evaluation set:

• Loss: 1.2144
• Accuracy: 0.7569
• Perplexity: 3.3682

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: 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 results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Perplexity
1.4325	1.0	14	1.2906	0.7533	3.6351
1.4038	2.0	28	1.2569	0.7560	3.5146
1.3795	3.0	42	1.2207	0.7535	3.3895
1.3515	4.0	56	1.1725	0.7632	3.2301
1.3173	5.0	70	1.1807	0.7625	3.2565
1.271	6.0	84	1.1901	0.7561	3.2874
1.2655	7.0	98	1.1605	0.7615	3.1915
1.2681	8.0	112	1.1396	0.7630	3.1255
1.2362	9.0	126	1.1118	0.7709	3.0397
1.2196	10.0	140	1.1419	0.7619	3.1328
1.2191	11.0	154	1.1508	0.7569	3.1606
1.1892	12.0	168	1.1398	0.7642	3.1263
1.1819	13.0	182	1.1509	0.7616	3.1612
1.1469	14.0	196	1.1664	0.7601	3.2103

Framework versions

• Transformers 4.48.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.2
• Tokenizers 0.21.0