vi-bert-base_v1

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

• Loss: 0.4995
• Accuracy: 0.9292
• Precision Macro: 0.8368
• Recall Macro: 0.7769
• F1 Macro: 0.8000
• F1 Weighted: 0.9259

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: 3e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• optimizer: Use 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
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision Macro	Recall Macro	F1 Macro	F1 Weighted
0.5604	1.0	90	0.2596	0.9128	0.9000	0.6666	0.6788	0.8972
0.2258	2.0	180	0.2182	0.9286	0.8216	0.8017	0.8109	0.9275
0.1532	3.0	270	0.2312	0.9198	0.7940	0.7902	0.7919	0.9195
0.123	4.0	360	0.2432	0.9311	0.8607	0.8000	0.8238	0.9286
0.0785	5.0	450	0.2592	0.9255	0.8450	0.7784	0.8037	0.9222
0.0628	6.0	540	0.3075	0.9280	0.8358	0.7765	0.7993	0.9247
0.0457	7.0	630	0.3155	0.9255	0.8118	0.7996	0.8053	0.9247
0.034	8.0	720	0.3924	0.9248	0.8212	0.7656	0.7870	0.9213
0.0271	9.0	810	0.3776	0.9242	0.8211	0.7782	0.7957	0.9216
0.0207	10.0	900	0.4209	0.9274	0.8067	0.8094	0.8080	0.9275
0.0189	11.0	990	0.4373	0.9255	0.7988	0.7957	0.7971	0.9252
0.0145	12.0	1080	0.4010	0.9349	0.8392	0.8228	0.8304	0.9341
0.0083	13.0	1170	0.4337	0.9242	0.8237	0.7988	0.8100	0.9228
0.004	14.0	1260	0.4571	0.9318	0.8491	0.7828	0.8080	0.9285
0.0081	15.0	1350	0.4862	0.9286	0.8298	0.7857	0.8035	0.9261
0.0027	16.0	1440	0.4788	0.9280	0.8348	0.7924	0.8103	0.9258
0.0029	17.0	1530	0.4797	0.9305	0.8339	0.7903	0.8085	0.9281
0.003	18.0	1620	0.4877	0.9280	0.8238	0.7807	0.7984	0.9253
0.0013	19.0	1710	0.4966	0.9286	0.8363	0.7765	0.7996	0.9253
0.0014	20.0	1800	0.4995	0.9292	0.8368	0.7769	0.8000	0.9259

Framework versions

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