KTB-finetune-wangchan2

This model is a fine-tuned version of airesearch/wangchanberta-base-att-spm-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.0846
• Accuracy: 0.4330
• Precision: 0.4624
• Recall: 0.4330
• F1: 0.3956

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 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	Precision	Recall	F1
No log	1.0	14	1.1169	0.3608	0.2185	0.3608	0.2722
No log	2.0	28	1.1113	0.3814	0.3102	0.3814	0.3303
No log	3.0	42	1.0851	0.4124	0.3384	0.4124	0.3373
No log	4.0	56	1.0999	0.4227	0.3430	0.4227	0.2880
No log	5.0	70	1.0953	0.3814	0.3760	0.3814	0.3100
No log	6.0	84	1.0783	0.4021	0.3109	0.4021	0.3262
No log	7.0	98	1.0914	0.3918	0.3012	0.3918	0.3218
No log	8.0	112	1.0998	0.3918	0.3676	0.3918	0.3396
No log	9.0	126	1.1017	0.4021	0.4020	0.4021	0.3461
No log	10.0	140	1.0662	0.4227	0.4291	0.4227	0.3448
No log	11.0	154	1.0933	0.4227	0.3507	0.4227	0.3444
No log	12.0	168	1.1042	0.3402	0.2775	0.3402	0.3029
No log	13.0	182	1.0863	0.4330	0.3640	0.4330	0.3515
No log	14.0	196	1.0963	0.4021	0.3690	0.4021	0.3332
No log	15.0	210	1.1130	0.4124	0.4155	0.4124	0.3500
No log	16.0	224	1.1093	0.4227	0.4300	0.4227	0.3810
No log	17.0	238	1.1093	0.4330	0.4527	0.4330	0.3885
No log	18.0	252	1.0891	0.4433	0.4751	0.4433	0.4092
No log	19.0	266	1.0850	0.4330	0.4624	0.4330	0.3956
No log	20.0	280	1.0846	0.4330	0.4624	0.4330	0.3956

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 2.14.4
• Tokenizers 0.21.1