phobert-base-v2_v3

This model is a fine-tuned version of vinai/phobert-base-v2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1699
• Accuracy: 0.7991
• Precision Macro: 0.7995
• Recall Macro: 0.7994
• F1 Macro: 0.7991
• F1 Weighted: 0.7990

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
1.086	1.0	72	0.7935	0.6887	0.6938	0.6887	0.6884	0.6882
0.8116	2.0	144	0.6396	0.7459	0.7595	0.7454	0.7452	0.7453
0.4754	3.0	216	0.5853	0.7796	0.7812	0.7801	0.7794	0.7793
0.3729	4.0	288	0.6464	0.7796	0.7827	0.7796	0.7795	0.7795
0.2206	5.0	360	0.6963	0.7845	0.7880	0.7842	0.7842	0.7843
0.161	6.0	432	0.7742	0.7827	0.7876	0.7829	0.7827	0.7826
0.1193	7.0	504	0.8773	0.7876	0.7906	0.7881	0.7873	0.7871
0.0858	8.0	576	0.9057	0.7800	0.7821	0.7797	0.7801	0.7801
0.081	9.0	648	0.9375	0.7871	0.7897	0.7875	0.7872	0.7871
0.0528	10.0	720	0.9239	0.7867	0.7880	0.7869	0.7868	0.7867
0.0459	11.0	792	0.9642	0.7920	0.7928	0.7920	0.7921	0.7921
0.0319	12.0	864	1.0246	0.7951	0.7960	0.7951	0.7952	0.7952
0.0303	13.0	936	1.0632	0.7960	0.7968	0.7963	0.7960	0.7958
0.0235	14.0	1008	1.0870	0.7956	0.7957	0.7955	0.7956	0.7955
0.0182	15.0	1080	1.1385	0.7960	0.7965	0.7963	0.7960	0.7959
0.0114	16.0	1152	1.1546	0.8013	0.8021	0.8012	0.8013	0.8013
0.0119	17.0	1224	1.1719	0.7969	0.7974	0.7972	0.7969	0.7968
0.0111	18.0	1296	1.1749	0.7960	0.7966	0.7963	0.7959	0.7958
0.0067	19.0	1368	1.1719	0.7991	0.7995	0.7994	0.7991	0.7990
0.0151	20.0	1440	1.1699	0.7991	0.7995	0.7994	0.7991	0.7990

Framework versions

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