metadata

library_name: transformers
license: mit
base_model: dbmdz/bert-base-turkish-cased
tags:
  - generated_from_trainer
model-index:
  - name: results
    results: []

results

This model is a fine-tuned version of dbmdz/bert-base-turkish-cased on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.3122
Hate Precision: 1.0
Hate Recall: 1.0
Hate F1: 1.0
Sexual Precision: 0.0
Sexual Recall: 0.0
Sexual F1: 0.0
Racism Precision: 0.0
Racism Recall: 0.0
Racism F1: 0.0
Threat Precision: 0.0
Threat Recall: 0.0
Threat F1: 0.0
Neutral Precision: 0.0
Neutral Recall: 0.0
Neutral F1: 0.0
Macro Precision: 0.2
Macro Recall: 0.2
Macro F1: 0.2

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: 16
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Hate Precision	Hate Recall	Hate F1	Macro Precision	Macro Recall	Macro F1
0.5808	5.0	10	0.4806	0.8333	1.0	0.9091	0.1667	0.2	0.1818
0.4203	10.0	20	0.3815	1.0	1.0	1.0	0.2	0.2	0.2
0.3519	15.0	30	0.3618	1.0	1.0	1.0	0.2	0.2	0.2
0.2748	20.0	40	0.3329	1.0	1.0	1.0	0.2	0.2	0.2
0.2373	25.0	50	0.3168	1.0	1.0	1.0	0.2	0.2	0.2
0.2129	30.0	60	0.3122	1.0	1.0	1.0	0.2	0.2	0.2

Framework versions

Transformers 4.45.1
Pytorch 2.4.0+cu124
Datasets 2.19.1
Tokenizers 0.20.0