cellate-tapt_base_ww_mask-LR_2e-05

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

• Loss: 1.3214
• Accuracy: 0.7147

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
1.3766	1.0	6	1.3727	0.7206
1.3628	2.0	12	1.3402	0.7168
1.3196	3.0	18	1.4073	0.7118
1.3119	4.0	24	1.3776	0.7206
1.3322	5.0	30	1.4455	0.7034
1.2673	6.0	36	1.4568	0.7093
1.2661	7.0	42	1.4778	0.6988
1.2646	8.0	48	1.3543	0.7289
1.2533	9.0	54	1.3255	0.7365
1.2382	10.0	60	1.4248	0.7088
1.2442	11.0	66	1.3088	0.7243
1.2318	12.0	72	1.3868	0.7126
1.2383	13.0	78	1.2669	0.7357
1.2914	14.0	84	1.3244	0.7264
1.2346	15.0	90	1.3439	0.7072
1.233	16.0	96	1.3672	0.7218
1.3158	16.7273	100	1.3214	0.7147

Framework versions

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