wav2vec2-large-xls-r-2b-georgian-v0.1

This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the common_voice dataset. It achieves the following results on the evaluation set:

• Loss: 0.2114
• Wer: 0.3321

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: 7e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 500
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.0674	8.16	400	1.4635	0.9765
1.6887	16.32	800	0.2869	0.4384
1.2844	24.48	1200	0.2669	0.3869
1.0713	32.64	1600	0.2244	0.3695
0.9464	40.8	2000	0.2126	0.3378
0.9069	48.96	2400	0.2114	0.3321

Framework versions

• Transformers 4.26.1
• Pytorch 1.10.0+cu113
• Datasets 2.9.0
• Tokenizers 0.13.2