whisper-tiny-multitask-hf-200k-ep10_practice_0805

This model is a fine-tuned version of openai/whisper-tiny on the arrow dataset. It achieves the following results on the evaluation set:

• Loss: 0.0082
• Cer: 100.0
• Acc: 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: 5e-05
• train_batch_size: 128
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Acc
0.54	1.0	1584	0.2232	100.0	1.0
0.1377	2.0	3168	0.0907	100.0	1.0
0.0538	3.0	4752	0.0473	100.0	1.0
0.0286	4.0	6336	0.0309	100.0	1.0
0.0178	5.0	7920	0.0214	100.0	1.0
0.011	6.0	9504	0.0167	100.0	1.0
0.008	7.0	11088	0.0130	100.0	1.0
0.0055	8.0	12672	0.0115	100.0	1.0
0.0038	9.0	14256	0.0094	100.0	1.0
0.0027	10.0	15840	0.0082	100.0	1.0

Framework versions

• Transformers 4.36.2
• Pytorch 2.0.1+cu117
• Datasets 2.16.1
• Tokenizers 0.15.0