bert-base-cased-ChemTok-ZN15-20KStat-V1

This model is a fine-tuned version of bert-base-cased on a subset of the cafierom/ZN1540K dataset of drug or drug-like molecules (20K molecules) with added tokens:

new_tokens = ["[C@H]","[C@@H]","(F)","(Cl)","c1","c2","(O)","N#C","(=O)",
              "([N+]([O-])=O)","[O-]","(OC)","(C)","[NH3+]","(I)","[Na+]","C#N"]

Model description

More information needed

Intended uses & limitations

It is meant to be used for finetuning classification models for drug-related tasks, and for generative unmasking.

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: 128
• eval_batch_size: 128
• 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: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss
2.8356	1.0	133	1.7723
1.3023	2.0	266	0.9532
0.9711	3.0	399	0.7969
0.8166	4.0	532	0.6598
0.6875	5.0	665	0.5524
0.6041	6.0	798	0.4795
0.5608	7.0	931	0.4678
0.5216	8.0	1064	0.4428
0.4963	9.0	1197	0.4071
0.4768	10.0	1330	0.4007
0.4601	11.0	1463	0.3836
0.4437	12.0	1596	0.3640
0.434	13.0	1729	0.3634
0.4226	14.0	1862	0.3517
0.4185	15.0	1995	0.3420
0.4107	16.0	2128	0.3460
0.3958	17.0	2261	0.3299
0.3943	18.0	2394	0.3413
0.3888	19.0	2527	0.3365
0.3926	20.0	2660	0.3312

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0