eraydikyologlu/bert_ayt_cografya

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:

• Train Loss: 0.1030
• Train Accuracy: 0.9857
• Validation Loss: 0.0405
• Validation Accuracy: 0.9963
• Epoch: 17

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:

• optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': None, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': True, 'is_legacy_optimizer': False, 'learning_rate': {'module': 'transformers.optimization_tf', 'class_name': 'WarmUp', 'config': {'initial_learning_rate': 2e-05, 'decay_schedule_fn': {'module': 'keras.optimizers.schedules', 'class_name': 'PolynomialDecay', 'config': {'initial_learning_rate': 2e-05, 'decay_steps': 7695, 'end_learning_rate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}, 'registered_name': None}, 'warmup_steps': 855, 'power': 1.0, 'name': None}, 'registered_name': 'WarmUp'}, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
• training_precision: float32

Training results

Train Loss	Train Accuracy	Validation Loss	Validation Accuracy	Epoch
4.0833	0.0676	3.5014	0.2831	0
2.7208	0.4063	2.0128	0.5009	1
1.8576	0.5336	1.4735	0.6140	2
1.4770	0.6020	1.2100	0.6572	3
1.2546	0.6466	1.0459	0.6811	4
1.0928	0.6801	0.8783	0.7243	5
0.9409	0.7235	0.7535	0.7638	6
0.8029	0.7618	0.5817	0.8364	7
0.6672	0.8104	0.4516	0.8824	8
0.5564	0.8459	0.3463	0.9173	9
0.4531	0.8843	0.2802	0.9311	10
0.3671	0.9062	0.2039	0.9522	11
0.2920	0.9319	0.1494	0.9715	12
0.2317	0.9512	0.1200	0.9816	13
0.1873	0.9633	0.0811	0.9890	14
0.1537	0.9717	0.0624	0.9908	15
0.1275	0.9799	0.0506	0.9945	16
0.1030	0.9857	0.0405	0.9963	17

Framework versions

• Transformers 4.52.4
• TensorFlow 2.18.0
• Datasets 2.14.4
• Tokenizers 0.21.1