phobert-large_nli

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

• Loss: 1.3062
• Accuracy: 0.8102
• Precision Macro: 0.8106
• Recall Macro: 0.8103
• F1 Macro: 0.8103
• F1 Weighted: 0.8103

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
1.0976	1.0	72	1.0257	0.5237	0.5529	0.5264	0.5082	0.5072
0.9271	2.0	144	0.6649	0.7592	0.7887	0.7579	0.7590	0.7590
0.4037	3.0	216	0.5864	0.7894	0.7930	0.7895	0.7895	0.7895
0.2866	4.0	288	0.6385	0.8120	0.8142	0.8125	0.8118	0.8118
0.1197	5.0	360	0.6949	0.8115	0.8117	0.8115	0.8115	0.8115
0.0939	6.0	432	0.7485	0.8058	0.8084	0.8060	0.8058	0.8059
0.0647	7.0	504	0.9244	0.7920	0.7977	0.7921	0.7919	0.7918
0.0457	8.0	576	0.8464	0.8106	0.8107	0.8107	0.8106	0.8106
0.046	9.0	648	0.9886	0.8062	0.8121	0.8066	0.8064	0.8063
0.026	10.0	720	0.9887	0.8120	0.8126	0.8121	0.8120	0.8121
0.0244	11.0	792	1.0642	0.8124	0.8130	0.8126	0.8125	0.8125
0.0211	12.0	864	1.0197	0.8075	0.8097	0.8078	0.8077	0.8077
0.0146	13.0	936	1.1487	0.8151	0.8171	0.8155	0.8151	0.8151
0.0085	14.0	1008	1.1846	0.8053	0.8056	0.8053	0.8053	0.8053
0.0051	15.0	1080	1.2905	0.8084	0.8095	0.8085	0.8084	0.8084
0.0036	16.0	1152	1.3259	0.8102	0.8121	0.8104	0.8104	0.8104
0.0027	17.0	1224	1.3187	0.8115	0.8121	0.8115	0.8116	0.8116
0.0023	18.0	1296	1.3024	0.8115	0.8120	0.8117	0.8116	0.8116
0.0025	19.0	1368	1.3049	0.8111	0.8115	0.8112	0.8111	0.8111
0.0037	20.0	1440	1.3062	0.8102	0.8106	0.8103	0.8103	0.8103

Framework versions

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