cellate-tapt_freeze_llrd-LR_2e-05

This model is a fine-tuned version of Mardiyyah/biomedbert_model_extended_untrained on the Mardiyyah/TAPT_CeLLaTe1.0 dataset. It achieves the following results on the evaluation set:

• Loss: 5.8892
• Accuracy: 0.2761

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 3407
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
10.1692	1.0	6	11.1264	0.0
9.9435	2.0	12	10.5601	0.0
9.3201	3.0	18	9.8066	0.0004
8.4869	4.0	24	8.8692	0.0080
7.8056	5.0	30	8.1258	0.0377
7.3126	6.0	36	7.8349	0.0633
6.9357	7.0	42	7.4240	0.0871
6.652	8.0	48	7.2423	0.1479
6.3972	9.0	54	6.9245	0.1860
6.2035	10.0	60	6.7783	0.2090
5.9674	11.0	66	6.5090	0.2233
5.8018	12.0	72	6.2905	0.2396
5.6325	13.0	78	6.1065	0.2698
5.581	14.0	84	6.0982	0.2694
5.4227	15.0	90	6.2502	0.2484
5.4131	16.0	96	5.9184	0.2669
5.726	16.7273	100	5.8348	0.2815

Framework versions

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