You can deploy this model and start earning money today!

lsnoo/
wav2vec2-large-xlsr-53k-russian

Name: wav2vec2-large-xlsr-53k-russian
Availability: InStock

transformers

pytorch

wav2vec2

automatic-speech-recognition

generated_from_trainer

common_voice

apache-2.0

endpoints_compatible

Views

No views yet

wav2vec2-large-xlsr-53k-russian

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the common_voice dataset. It achieves the following results on the evaluation set:

Loss: 0.2660
Wer: 0.2052

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: 96
eval_batch_size: 32
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 192
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 30
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.2873	1.09	400	0.8580	0.8982
0.4728	2.19	800	0.3182	0.3892
0.1639	9.83	1200	0.2374	0.2646
0.1014	13.11	1600	0.2470	0.2467
0.0754	16.39	2000	0.2516	0.2337
0.0616	19.67	2400	0.2559	0.2237
0.0505	22.95	2800	0.2557	0.2155
0.0437	26.23	3200	0.2711	0.2099
0.0377	29.51	3600	0.2660	0.2052

Framework versions

Transformers 4.17.0
Pytorch 1.11.0+cu113
Datasets 1.18.3
Tokenizers 0.12.1

lsnoo/wav2vec2-large-xlsr-53k-russian

wav2vec2-large-xlsr-53k-russian

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

lsnoo/
wav2vec2-large-xlsr-53k-russian