wav2vec2-large-xlsr-53-AsanteTwi-05

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the common_voice_13_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.7657
• Wer: 0.75

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• num_epochs: 150

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
2.2241	16.67	100	2.1317	1.0
1.5168	33.33	200	1.1019	0.8125
0.7964	50.0	300	0.7658	0.75
0.4985	66.67	400	0.6807	0.625
0.3885	83.33	500	0.7197	0.5625
0.3269	100.0	600	0.7616	0.5625
0.2625	116.67	700	0.7000	0.6875
0.2595	133.33	800	0.7425	0.6875
0.2388	150.0	900	0.7657	0.75

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.1+cu118
• Datasets 2.13.1
• Tokenizers 0.13.3