This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice_13_0 dataset. It achieves the following results on the evaluation set:

Loss: 0.2513
Wer: 0.2459

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: 1
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 512
total_train_batch_size: 512
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
5.4424	4.88	400	0.4321	0.4851
0.2599	9.77	800	0.2841	0.3252
0.1511	14.65	1200	0.2664	0.2900
0.1124	19.54	1600	0.2671	0.2687
0.0869	24.42	2000	0.2531	0.2586
0.068	29.31	2400	0.2513	0.2459

Framework versions

Transformers 4.35.0.dev0
Pytorch 1.10.0+cu113
Datasets 2.14.6
Tokenizers 0.14.1

MimionanA/wav2vec2-large-xls-r-300m-turkish-colab

wav2vec2-large-xls-r-300m-turkish-colab

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

MimionanA/
wav2vec2-large-xls-r-300m-turkish-colab