Gummybear05/whisper-small-E30_speed2

whisper-small-E30_Yspeed

This model is a fine-tuned version of openai/whisper-small on the aihub old adult freq speed pause changed dataset. It achieves the following results on the evaluation set:

• Loss: 0.1835
• Cer: 4.8520

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 50
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.2793	0.1289	100	0.2591	6.6318
0.1561	0.2579	200	0.2254	6.1560
0.1532	0.3868	300	0.2258	6.3205
0.1394	0.5158	400	0.2110	5.7096
0.1102	0.6447	500	0.2068	5.6391
0.1085	0.7737	600	0.2006	5.4159
0.108	0.9026	700	0.1983	5.5392
0.0547	1.0309	800	0.1861	5.1046
0.0462	1.1599	900	0.1899	4.9225
0.045	1.2888	1000	0.1920	5.0399
0.0407	1.4178	1100	0.1881	4.8167
0.0391	1.5467	1200	0.1845	4.7227
0.0403	1.6757	1300	0.1855	5.1222
0.0378	1.8046	1400	0.1846	4.8461
0.041	1.9336	1500	0.1835	4.8520

Framework versions

• Transformers 4.47.0.dev0
• Pytorch 2.5.0+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3