review_classification_bert_base_jp_v3_ratio1_1_2label_add_dropout-epoch40_v4

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: 1.4594
  • Accuracy: 0.8023

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: 1e-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: 40

Training results

Training Loss Epoch Step Validation Loss Accuracy
No log 1.0 52 0.5181 0.7316
No log 2.0 104 0.4858 0.7684
No log 3.0 156 0.4934 0.7684
No log 4.0 208 0.5022 0.7910
No log 5.0 260 0.5365 0.7910
No log 6.0 312 0.6070 0.8136
No log 7.0 364 0.7252 0.8107
No log 8.0 416 0.9174 0.7825
No log 9.0 468 1.0906 0.7655
0.208 10.0 520 1.2154 0.7740
0.208 11.0 572 1.0851 0.8051
0.208 12.0 624 1.3699 0.7627
0.208 13.0 676 1.1715 0.7994
0.208 14.0 728 1.2354 0.7910
0.208 15.0 780 1.2615 0.7881
0.208 16.0 832 1.2891 0.7825
0.208 17.0 884 1.4250 0.7768
0.208 18.0 936 1.5752 0.7712
0.208 19.0 988 1.3456 0.8023
0.001 20.0 1040 1.3598 0.7994
0.001 21.0 1092 1.3378 0.8107
0.001 22.0 1144 1.3534 0.8051
0.001 23.0 1196 1.3832 0.7881
0.001 24.0 1248 1.3838 0.7938
0.001 25.0 1300 1.4257 0.7881
0.001 26.0 1352 1.4115 0.8023
0.001 27.0 1404 1.6230 0.7684
0.001 28.0 1456 1.4132 0.8023
0.0009 29.0 1508 1.6508 0.7684
0.0009 30.0 1560 1.4539 0.8023
0.0009 31.0 1612 1.4677 0.8023
0.0009 32.0 1664 1.4155 0.8079
0.0009 33.0 1716 1.4262 0.8079
0.0009 34.0 1768 1.4390 0.8023
0.0009 35.0 1820 1.4405 0.8023
0.0009 36.0 1872 1.4471 0.8051
0.0009 37.0 1924 1.4526 0.8023
0.0009 38.0 1976 1.4568 0.8023
0.0001 39.0 2028 1.4596 0.8023
0.0001 40.0 2080 1.4594 0.8023

Framework versions

  • Transformers 4.35.2
  • Pytorch 2.1.0+cu121
  • Datasets 2.15.0
  • Tokenizers 0.15.0
Downloads last month
4
Safetensors
Model size
111M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported third-party Inference Providers, and the model is not deployed on the HF Inference API.

Model tree for kt220/review_classification_bert_base_jp_v3_ratio1_1_2label_add_dropout-epoch40_v4

Finetuned
(36)
this model