review_classification_bert_base_jp_v3_ratio1_1-epoch20

This model is a fine-tuned version of cl-tohoku/bert-base-japanese-v3 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.5045
• Accuracy: 0.6888

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	85	0.5663	0.7080
No log	2.0	170	0.6268	0.6947
No log	3.0	255	0.7858	0.6888
No log	4.0	340	0.9156	0.6962
No log	5.0	425	1.2999	0.6792
0.2773	6.0	510	1.9478	0.6600
0.2773	7.0	595	2.0691	0.6733
0.2773	8.0	680	2.0470	0.6755
0.2773	9.0	765	2.0596	0.7050
0.2773	10.0	850	2.1844	0.6954
0.2773	11.0	935	2.3059	0.6858
0.0102	12.0	1020	2.4167	0.6792
0.0102	13.0	1105	2.4146	0.6881
0.0102	14.0	1190	2.3954	0.6947
0.0102	15.0	1275	2.3927	0.6954
0.0102	16.0	1360	2.4824	0.6858
0.0102	17.0	1445	2.4586	0.6954
0.0011	18.0	1530	2.4730	0.6947
0.0011	19.0	1615	2.5057	0.6873
0.0011	20.0	1700	2.5045	0.6888

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0