CheeLi03/whisper-base-rus-8

Whisper Base Russian 8000 - Chee Li

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

• Loss: 0.4957
• Wer: 25.5545

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: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 850
• training_steps: 8000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0635	5.4645	1000	0.3433	22.5882
0.0051	10.9290	2000	0.3879	23.0492
0.0019	16.3934	3000	0.4186	23.8976
0.0011	21.8579	4000	0.4422	24.4522
0.0007	27.3224	5000	0.4613	25.0
0.0005	32.7869	6000	0.4781	25.3140
0.0004	38.2514	7000	0.4907	25.4209
0.0003	43.7158	8000	0.4957	25.5545

Framework versions

• Transformers 4.43.4
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1