bert-base-cased-ChemTok-ZN15-55KTyrosinase-V1-finetuned-Tyrosinase-IC50s-V4

This model is a fine-tuned version of cafierom/bert-base-cased-ChemTok-ZN15-55KTyrosinase-V1 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1854
• Accuracy: 0.7028
• F1: 0.6998

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: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.0678	1.0	19	1.0197	0.4953	0.4712
0.9633	2.0	38	0.9236	0.5425	0.5418
0.896	3.0	57	0.9251	0.5566	0.5502
0.8178	4.0	76	0.9288	0.5660	0.5506
0.7743	5.0	95	0.8277	0.6085	0.6115
0.6971	6.0	114	0.8013	0.6651	0.6669
0.6214	7.0	133	0.8409	0.6179	0.6195
0.5886	8.0	152	0.8234	0.6651	0.6627
0.5337	9.0	171	0.8363	0.6792	0.6800
0.4705	10.0	190	0.8959	0.6745	0.6744
0.4393	11.0	209	0.9193	0.6792	0.6810
0.4089	12.0	228	0.9230	0.6887	0.6877
0.3926	13.0	247	0.9601	0.6934	0.6941
0.3897	14.0	266	0.9610	0.6981	0.6987
0.326	15.0	285	0.9510	0.7075	0.7085
0.3214	16.0	304	1.0061	0.6792	0.6712
0.3317	17.0	323	1.0278	0.6840	0.6799
0.2843	18.0	342	1.0050	0.6981	0.6971
0.2409	19.0	361	1.0434	0.7075	0.7088
0.2247	20.0	380	1.1854	0.7028	0.6998

Framework versions

• Transformers 4.50.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1