phobert-base_v3

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

• Loss: 1.2798
• Accuracy: 0.7805
• Precision Macro: 0.7813
• Recall Macro: 0.7807
• F1 Macro: 0.7806
• F1 Weighted: 0.7806

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: 5e-05
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 256
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision Macro	Recall Macro	F1 Macro	F1 Weighted
1.0861	1.0	72	0.8945	0.5863	0.6024	0.5883	0.5734	0.5725
0.8854	2.0	144	0.7401	0.6993	0.7354	0.6992	0.6982	0.6983
0.5405	3.0	216	0.5891	0.7814	0.7817	0.7816	0.7813	0.7813
0.4119	4.0	288	0.6523	0.7761	0.7776	0.7758	0.7760	0.7760
0.2355	5.0	360	0.6712	0.7894	0.7899	0.7892	0.7894	0.7894
0.1786	6.0	432	0.8116	0.7725	0.7733	0.7726	0.7726	0.7726
0.1126	7.0	504	0.8907	0.7761	0.7792	0.7761	0.7761	0.7761
0.0844	8.0	576	0.9184	0.7827	0.7834	0.7825	0.7826	0.7827
0.0657	9.0	648	1.0276	0.7734	0.7769	0.7735	0.7737	0.7737
0.0458	10.0	720	1.2265	0.7583	0.7713	0.7581	0.7582	0.7583
0.0494	11.0	792	1.1001	0.7783	0.7793	0.7783	0.7784	0.7784
0.0307	12.0	864	1.1487	0.7783	0.7798	0.7781	0.7783	0.7783
0.0284	13.0	936	1.1877	0.7805	0.7812	0.7805	0.7805	0.7805
0.0192	14.0	1008	1.2280	0.7836	0.7843	0.7839	0.7836	0.7836
0.0172	15.0	1080	1.2466	0.7823	0.7823	0.7823	0.7823	0.7823
0.0108	16.0	1152	1.2673	0.7809	0.7837	0.7810	0.7812	0.7812
0.0111	17.0	1224	1.2614	0.7823	0.7825	0.7823	0.7823	0.7823
0.0094	18.0	1296	1.2754	0.7814	0.7817	0.7815	0.7814	0.7815
0.0079	19.0	1368	1.2823	0.7809	0.7821	0.7811	0.7811	0.7811
0.0095	20.0	1440	1.2798	0.7805	0.7813	0.7807	0.7806	0.7806

Framework versions

• Transformers 4.55.0
• Pytorch 2.7.0+cu126
• Datasets 4.0.0
• Tokenizers 0.21.4