cellate2.0-tapt_base-LR_5e-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: 1.1527
• Accuracy: 0.7405

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: 5e-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: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.385	1.0	4	1.3146	0.7350
1.4161	2.0	8	1.2337	0.7449
1.3044	3.0	12	1.2215	0.7331
1.3033	4.0	16	1.3410	0.7153
1.2797	5.0	20	1.2776	0.7228
1.265	6.0	24	1.2818	0.7273
1.2078	7.0	28	1.2216	0.7442
1.1611	8.0	32	1.3203	0.7321
1.1928	9.0	36	1.2691	0.7335
1.1883	10.0	40	1.2359	0.7408
1.1549	11.0	44	1.2654	0.7350
1.1768	12.0	48	1.2919	0.7224
1.1169	13.0	52	1.2514	0.7263
1.1635	14.0	56	1.2352	0.7324
1.0975	15.0	60	1.2906	0.7385
1.0897	16.0	64	1.2099	0.7337
0.9993	17.0	68	1.2057	0.7395
1.0331	18.0	72	1.2895	0.7347
1.1321	19.0	76	1.2824	0.7386
1.083	20.0	80	1.1527	0.7405

Framework versions

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