wav2vec2-large-xls-r-300m-pashto-colab-test-6

This model is a fine-tuned version of rsd16/wav2vec2-large-xlsr-53-fine-tuned-farsi on the audiofolder dataset. It achieves the following results on the evaluation set:

• Loss: nan
• Wer: 1.0

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: 0.9
• train_batch_size: 1
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 10
• total_train_batch_size: 10
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1860014104903.68	0.96	100	nan	1.0
0.0	1.91	200	nan	1.0
0.0	2.87	300	nan	1.0
0.0	3.82	400	nan	1.0
0.0	4.78	500	nan	1.0
0.0	5.73	600	nan	1.0
0.0	6.69	700	nan	1.0
0.0	7.64	800	nan	1.0
0.0	8.6	900	nan	1.0
0.0	9.55	1000	nan	1.0
0.0	10.51	1100	nan	1.0
0.0	11.46	1200	nan	1.0
0.0	12.42	1300	nan	1.0
0.0	13.37	1400	nan	1.0
0.0	14.33	1500	nan	1.0

Framework versions

• Transformers 4.29.2
• Pytorch 2.0.1+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3