whisper-base-full-data-language-v2-20ep

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

• Loss: 0.1929

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: 0.00015
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• distributed_type: tpu
• num_devices: 8
• gradient_accumulation_steps: 2
• 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
0.3116	1.57	5000	0.5301
0.2104	3.13	10000	0.4066
0.1729	4.7	15000	0.3555
0.1472	6.27	20000	0.3208
0.128	7.83	25000	0.2923
0.1065	9.4	30000	0.2719
0.0995	10.97	35000	0.2516
0.0812	12.53	40000	0.2368
0.066	14.1	45000	0.2230
0.0574	15.67	50000	0.2119
0.0463	17.23	55000	0.2028
0.04	18.8	60000	0.1957

Framework versions

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