whisper-tiny-atcosim

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

• Loss: 0.0711
• Wer: 72.8237

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: 128
• eval_batch_size: 128
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2141	8.33	500	0.0633	15.6047
0.0023	16.67	1000	0.0629	29.2091
0.0007	25.0	1500	0.0646	46.2076
0.0003	33.33	2000	0.0659	54.1767
0.0002	41.67	2500	0.0670	58.2284
0.0002	50.0	3000	0.0679	64.0952
0.0001	58.33	3500	0.0688	65.9520
0.0001	66.67	4000	0.0695	68.5081
0.0001	75.0	4500	0.0701	70.5316
0.0001	83.33	5000	0.0706	72.2217
0.0001	91.67	5500	0.0710	72.6801
0.0001	100.0	6000	0.0711	72.8237

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2
• Datasets 2.15.0
• Tokenizers 0.15.0