This time use self-made dataset(cut the audio of "https://www.youtube.com/watch?v=a2ZOTD3R7JI" into slices and write the corresponding transcript, totally 4 mins), don't know why the word-error-rate keep 1. But can know that much be the problem of dataset, because last time use the same pre-trained model and standard singlish corpus fine-tune get nice result. (can find it at:RuiqianLi/wav2vec2-large-xls-r-300m-singlish-colab)
It achieves the following results on the evaluation set:
Loss: 3.0927
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: 0.01
train_batch_size: 2
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 4
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
