CeLLaTe-tapt-biofomer16l-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: 1.3781
• Accuracy: 0.7283
• Perplexity: 3.9672

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	0.7111	5.0472
1.8799	2.0	30	1.5831	0.7163	4.8698
1.801	3.0	45	1.5290	0.7222	4.6138
1.7444	4.0	60	1.4856	0.7248	4.4177
1.7086	5.0	75	1.4688	0.7270	4.3441
1.6495	6.0	90	1.4739	0.7238	4.3661
1.6387	7.0	105	1.4182	0.7269	4.1299
1.5975	8.0	120	1.4524	0.7229	4.2735
1.5862	9.0	135	1.3966	0.7306	4.0413
1.5551	10.0	150	1.3948	0.7332	4.0344
1.5185	11.0	165	1.3779	0.7324	3.9666
1.508	12.0	180	1.3371	0.7407	3.8080
1.4904	13.0	195	1.4013	0.7317	4.0606
1.4741	14.0	210	1.3646	0.7361	3.9143
1.458	15.0	225	1.3648	0.7342	3.9150
1.4591	16.0	240	1.3683	0.7310	3.9286
1.432	17.0	255	1.3534	0.7345	3.8704

Framework versions

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