You can deploy this model and start earning money today!

utakumi/
Hubert-noisy-cv-kakeiken-G

Name: Hubert-noisy-cv-kakeiken-G
Rating: 2.5 (5 reviews)

transformers

safetensors

hubert

automatic-speech-recognition

original_noisy_kakeiken_A

generated_from_trainer

rinna/japanese-hubert-base

finetune

apache-2.0

endpoints_compatible

Views

No views yet

Hubert-noisy-cv-kakeiken-G

This model is a fine-tuned version of rinna/japanese-hubert-base on the ORIGINAL_NOISY_KAKEIKEN_A - JA dataset. It achieves the following results on the evaluation set:

Loss: 0.0076
Wer: 0.9994
Cer: 0.0885

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: 0.0003
train_batch_size: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
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_steps: 12500
num_epochs: 5.0
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.1785	1.0	2732	0.0648	0.9996	0.0994
0.0328	2.0	5464	0.0264	0.9995	0.0946
0.0241	3.0	8196	0.0281	0.9997	0.0940
0.0288	4.0	10928	0.0323	0.9996	0.0939
0.0126	4.9984	13655	0.0076	0.9994	0.0885

Framework versions

Transformers 4.48.0
Pytorch 2.5.1+cu124
Datasets 3.1.0
Tokenizers 0.21.0

utakumi/Hubert-noisy-cv-kakeiken-G

Hubert-noisy-cv-kakeiken-G

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

utakumi/
Hubert-noisy-cv-kakeiken-G