whisper-small-full-data-language-v1-20ep

This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1491

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: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: tpu
• num_devices: 8
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• total_eval_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 5000
• training_steps: 63840

Training results

Training Loss	Epoch	Step	Validation Loss
0.1819	1.57	5000	0.3275
0.1222	3.13	10000	0.2598
0.1007	4.7	15000	0.2338
0.0847	6.27	20000	0.2120
0.0743	7.83	25000	0.2002
0.0601	9.4	30000	0.1898
0.0561	10.97	35000	0.1776
0.0441	12.53	40000	0.1712
0.0359	14.1	45000	0.1652
0.0303	15.67	50000	0.1583
0.0239	17.23	55000	0.1543
0.0206	18.8	60000	0.1504

Framework versions

• Transformers 4.28.1
• Pytorch 2.1.0a0+gitcc01568
• Datasets 2.13.1
• Tokenizers 0.13.3