whisper-tiny-el-lora

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

• Loss: 0.7610
• Wer: 0.7897

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: 2
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• 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
• training_steps: 5000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.9613	5.9668	250	0.9019	0.8102
0.8925	11.9184	500	0.8434	0.8005
0.8578	17.8701	750	0.8242	0.8119
0.8528	23.8218	1000	0.8114	0.8133
0.8409	29.7734	1250	0.8012	0.8001
0.84	35.7251	1500	0.7932	0.8136
0.8285	41.6767	1750	0.7874	0.7946
0.8293	47.6284	2000	0.7822	0.7970
0.7985	53.5801	2250	0.7782	0.8227
0.7834	59.5317	2500	0.7748	0.8136
0.7673	65.4834	2750	0.7717	0.7994
0.7803	71.4350	3000	0.7692	0.8046
0.7774	77.3867	3250	0.7668	0.8060
0.8034	83.3384	3500	0.7654	0.8057
0.796	89.2900	3750	0.7641	0.8060
0.7909	95.2417	4000	0.7632	0.7922
0.7524	101.1934	4250	0.7618	0.7908
0.8008	107.1450	4500	0.7612	0.7901
0.8008	113.0967	4750	0.7610	0.7901
0.7678	119.0483	5000	0.7610	0.7897

Framework versions

• PEFT 0.15.2
• Transformers 4.52.3
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1