vit5-base_nli

This model is a fine-tuned version of VietAI/vit5-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.9361
• Accuracy: 0.7508
• Precision Macro: 0.7512
• Recall Macro: 0.7507
• F1 Macro: 0.7508
• F1 Weighted: 0.7508

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
1.2042	1.0	72	1.0168	0.4772	0.5084	0.4784	0.4539	0.4536
1.0195	2.0	144	0.7645	0.6723	0.6767	0.6724	0.6711	0.6709
0.6388	3.0	216	0.6869	0.7126	0.7199	0.7121	0.7128	0.7127
0.5149	4.0	288	0.6967	0.7428	0.7453	0.7432	0.7427	0.7425
0.2882	5.0	360	0.7899	0.7375	0.7440	0.7375	0.7376	0.7374
0.2238	6.0	432	0.9740	0.7313	0.7398	0.7319	0.7300	0.7298
0.1326	7.0	504	1.0921	0.7344	0.7372	0.7350	0.7337	0.7335
0.096	8.0	576	1.2234	0.7366	0.7420	0.7361	0.7366	0.7366
0.0755	9.0	648	1.3014	0.7326	0.7355	0.7324	0.7332	0.7330
0.0505	10.0	720	1.3717	0.7397	0.7414	0.7395	0.7400	0.7399
0.0419	11.0	792	1.4521	0.7392	0.7429	0.7389	0.7394	0.7393
0.0301	12.0	864	1.5602	0.7428	0.7433	0.7428	0.7430	0.7429
0.0213	13.0	936	1.7194	0.7450	0.7457	0.7448	0.7450	0.7450
0.0171	14.0	1008	1.7975	0.7450	0.7475	0.7448	0.7449	0.7449
0.018	15.0	1080	1.7963	0.7525	0.7528	0.7525	0.7526	0.7526
0.0084	16.0	1152	1.8312	0.7512	0.7517	0.7512	0.7513	0.7513
0.0083	17.0	1224	1.8834	0.7525	0.7531	0.7526	0.7526	0.7525
0.0089	18.0	1296	1.9212	0.7561	0.7568	0.7561	0.7562	0.7561
0.0064	19.0	1368	1.9379	0.7508	0.7512	0.7507	0.7508	0.7508
0.0082	20.0	1440	1.9361	0.7508	0.7512	0.7507	0.7508	0.7508

Framework versions

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