wav2vec2-XLS-R-Fleurs-demo-google-colab-Ezra_William_Prod7

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

• Loss: 1.0673
• Wer: 0.5033

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: 0.001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 600
• num_epochs: 90
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
5.2829	7.79	300	2.8538	1.0
1.9733	15.58	600	0.8923	0.7851
0.4186	23.38	900	0.8297	0.6443
0.2077	31.17	1200	0.8573	0.6011
0.1535	38.96	1500	0.9490	0.5800
0.1163	46.75	1800	1.0380	0.5652
0.1001	54.55	2100	0.9354	0.5417
0.0845	62.34	2400	1.0226	0.5364
0.0711	70.13	2700	1.0799	0.5220
0.0588	77.92	3000	1.0550	0.5050
0.0492	85.71	3300	1.0673	0.5033

Framework versions

• Transformers 4.39.0
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2