hubert-rinnna-jp-jdrtsp-fw07sp-12

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

• Loss: 1.1989
• Wer: 0.6801
• Cer: 0.5794

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: 32
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
5.0318	1.0	404	4.2999	0.9798	0.9889
3.5113	2.0	808	3.3289	0.9798	0.9889
2.7536	3.0	1212	2.7007	0.9798	0.9889
2.4826	4.0	1616	2.3732	0.9798	0.9889
2.0642	5.0	2020	1.9165	0.9798	0.9888
1.834	6.0	2424	1.6739	0.9504	0.9464
1.6869	7.0	2828	1.4651	0.8239	0.7865
1.5734	8.0	3232	1.3267	0.7440	0.6939
1.5052	9.0	3636	1.2331	0.7045	0.6231
1.4573	10.0	4040	1.1989	0.6801	0.5794

Framework versions

• Transformers 4.34.0.dev0
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3