Sabanci-IT-Destek-v2

This model is a fine-tuned version of dbmdz/bert-base-turkish-cased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.1084
• Accuracy: 0.8131
• F1: 0.8114

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: Use OptimizerNames.ADAMW_TORCH_FUSED 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	Accuracy	F1
No log	1.0	422	4.0712	0.0383	0.0360
4.2799	2.0	844	3.3001	0.3098	0.2829
3.6351	3.0	1266	2.3719	0.5754	0.5485
2.6628	4.0	1688	1.7979	0.6600	0.6509
1.8814	5.0	2110	1.5134	0.7042	0.6993
1.3619	6.0	2532	1.2919	0.7403	0.7257
1.3619	7.0	2954	1.1487	0.7344	0.7301
0.9968	8.0	3376	1.1233	0.7609	0.7568
0.7387	9.0	3798	1.0679	0.7712	0.7688
0.5883	10.0	4220	1.0569	0.7763	0.7680
0.4789	11.0	4642	1.0722	0.7873	0.7802
0.4034	12.0	5064	1.0593	0.7918	0.7883
0.4034	13.0	5486	1.0518	0.7859	0.7792
0.3146	14.0	5908	1.0920	0.7910	0.7851
0.2744	15.0	6330	1.0781	0.8035	0.7984
0.243	16.0	6752	1.0820	0.8072	0.8010
0.2105	17.0	7174	1.1077	0.8065	0.8030
0.1966	18.0	7596	1.1005	0.8138	0.8110
0.1774	19.0	8018	1.1022	0.8138	0.8114
0.1774	20.0	8440	1.1084	0.8131	0.8114

Framework versions

• Transformers 4.57.3
• Pytorch 2.9.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.2