Edit model card

prot_bert_classification_finetuned_karolina_es_20e

This model is a fine-tuned version of nepp1d0/prot_bert-finetuned-smiles-bindingDB on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6763
  • Accuracy: 0.92
  • F1: 0.9583
  • Precision: 1.0
  • Recall: 0.92

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: 1e-06
  • train_batch_size: 64
  • eval_batch_size: 64
  • seed: 3
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 100
  • num_epochs: 20

Training results

Training Loss Epoch Step Validation Loss Accuracy F1 Precision Recall
No log 1.0 2 0.7084 0.02 0.0392 1.0 0.02
No log 2.0 4 0.7082 0.02 0.0392 1.0 0.02
No log 3.0 6 0.7078 0.04 0.0769 1.0 0.04
No log 4.0 8 0.7072 0.04 0.0769 1.0 0.04
No log 5.0 10 0.7065 0.04 0.0769 1.0 0.04
No log 6.0 12 0.7055 0.04 0.0769 1.0 0.04
No log 7.0 14 0.7044 0.04 0.0769 1.0 0.04
No log 8.0 16 0.7031 0.06 0.1132 1.0 0.06
No log 9.0 18 0.7017 0.12 0.2143 1.0 0.12
No log 10.0 20 0.6999 0.2 0.3333 1.0 0.2
No log 11.0 22 0.6981 0.22 0.3607 1.0 0.22
No log 12.0 24 0.6962 0.22 0.3607 1.0 0.22
No log 13.0 26 0.6941 0.24 0.3871 1.0 0.24
No log 14.0 28 0.6917 0.44 0.6111 1.0 0.44
No log 15.0 30 0.6893 0.58 0.7342 1.0 0.58
No log 16.0 32 0.6869 0.76 0.8636 1.0 0.76
No log 17.0 34 0.6842 0.88 0.9362 1.0 0.88
No log 18.0 36 0.6816 0.9 0.9474 1.0 0.9
No log 19.0 38 0.6789 0.92 0.9583 1.0 0.92
No log 20.0 40 0.6763 0.92 0.9583 1.0 0.92

Framework versions

  • Transformers 4.23.1
  • Pytorch 1.11.0
  • Datasets 2.6.1
  • Tokenizers 0.13.1
Downloads last month
8
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.