CeLLaTe-tapt-bioformer16l-tokenizer-adapted-baseline

This model is a fine-tuned version of bioformers/bioformer-16l on the Mardiyyah/TAPT_CeLLaTe2.0 dataset. It achieves the following results on the evaluation set:

Loss: 1.3768
Accuracy: 0.7284
Perplexity: 3.9623

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
1.9156	1.0	15	1.6188	0.7112	5.0472
1.8799	2.0	30	1.5831	0.7161	4.8702
1.8013	3.0	45	1.5291	0.7222	4.6141
1.7449	4.0	60	1.4859	0.7253	4.4191
1.7088	5.0	75	1.4696	0.7265	4.3475
1.6504	6.0	90	1.4740	0.7237	4.3665
1.6389	7.0	105	1.4179	0.7270	4.1284
1.5959	8.0	120	1.4521	0.7231	4.2721
1.5839	9.0	135	1.3957	0.7304	4.0378
1.5529	10.0	150	1.3926	0.7328	4.0254
1.5166	11.0	165	1.3766	0.7325	3.9613
1.5055	12.0	180	1.3351	0.7398	3.8003
1.4887	13.0	195	1.4013	0.7315	4.0604
1.4724	14.0	210	1.3644	0.7362	3.9134
1.4553	15.0	225	1.3633	0.7352	3.9093
1.4581	16.0	240	1.3672	0.7312	3.9245
1.4307	17.0	255	1.3518	0.7351	3.8645