<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
bert-finetuned-bpmn
This model is a fine-tuned version of bert-base-cased on a dataset containing textual process descriptions.
The dataset contains 2 target labels:
AGENTTASK
The dataset (and the notebook used for training) can be found on the following GitHub repo: https://github.com/jtlicardo/bert-finetuned-bpmn
Update: a model trained on 5 BPMN-specific labels can be found here: https://huggingface.co/jtlicardo/bpmn-information-extraction
The model achieves the following results on the evaluation set:
- Loss: 0.2656
- Precision: 0.7314
- Recall: 0.8366
- F1: 0.7805
- Accuracy: 0.8939
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: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| No log | 1.0 | 10 | 0.8437 | 0.1899 | 0.3203 | 0.2384 | 0.7005 |
| No log | 2.0 | 20 | 0.4967 | 0.5421 | 0.7582 | 0.6322 | 0.8417 |
| No log | 3.0 | 30 | 0.3403 | 0.6719 | 0.8431 | 0.7478 | 0.8867 |
| No log | 4.0 | 40 | 0.2821 | 0.6923 | 0.8235 | 0.7522 | 0.8903 |
| No log | 5.0 | 50 | 0.2656 | 0.7314 | 0.8366 | 0.7805 | 0.8939 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu116
- Datasets 2.7.1
- Tokenizers 0.13.2
