whisper-tiny-fi-lora

This model is a fine-tuned version of openai/whisper-tiny on the common_voice_11_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.6958
• Wer: 75.1203

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
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.7258	0.3690	100	3.6261	82.2153
2.1228	0.7380	200	1.9384	86.0270
0.951	1.1070	300	0.9630	82.1964
0.7932	1.4760	400	0.8430	82.5644
0.7635	1.8450	500	0.8000	85.1967
0.7242	2.2140	600	0.7735	79.3094
0.7099	2.5830	700	0.7573	82.7437
0.6686	2.9520	800	0.7504	80.5265
0.6476	3.3210	900	0.7415	78.7716
0.6494	3.6900	1000	0.7316	83.7626
0.6069	4.0590	1100	0.7307	78.5263
0.6463	4.4280	1200	0.7254	79.0358
0.5897	4.7970	1300	0.7210	78.9414
0.5816	5.1661	1400	0.7161	79.0924
0.5677	5.5351	1500	0.7174	76.4978
0.5584	5.9041	1600	0.7116	77.7715
0.5027	6.2731	1700	0.7081	76.0921
0.5214	6.6421	1800	0.7114	76.3657
0.5503	7.0111	1900	0.7113	76.3751
0.5057	7.3801	2000	0.7065	75.7713
0.5338	7.7491	2100	0.7052	76.4978
0.4457	8.1181	2200	0.7052	75.8562
0.5183	8.4871	2300	0.7017	76.7337
0.4988	8.8561	2400	0.7006	75.9600
0.4858	9.2251	2500	0.7001	75.6958
0.5024	9.5941	2600	0.7009	76.8752
0.5111	9.9631	2700	0.6998	75.6015
0.4985	10.3321	2800	0.6987	77.9791
0.4725	10.7011	2900	0.6975	77.4035
0.4497	11.0701	3000	0.6970	75.3090
0.4534	11.4391	3100	0.6972	75.4883
0.4839	11.8081	3200	0.6962	78.0262
0.4543	12.1771	3300	0.6970	75.7147
0.4586	12.5461	3400	0.6978	75.6581
0.4656	12.9151	3500	0.6997	76.3374
0.4177	13.2841	3600	0.6951	76.0449
0.4443	13.6531	3700	0.6965	75.3279
0.4698	14.0221	3800	0.6975	75.3562
0.4412	14.3911	3900	0.6957	75.2807
0.4027	14.7601	4000	0.6955	77.0356
0.4755	15.1292	4100	0.6963	75.4505
0.4487	15.4982	4200	0.6950	75.1014
0.4237	15.8672	4300	0.6967	75.2241
0.4222	16.2362	4400	0.6975	75.3090
0.408	16.6052	4500	0.6975	75.2618
0.4671	16.9742	4600	0.6947	75.0448
0.448	17.3432	4700	0.6954	75.2901
0.4253	17.7122	4800	0.6956	75.0826
0.44	18.0812	4900	0.6959	75.1392
0.4053	18.4502	5000	0.6958	75.1203

Framework versions

• Transformers 4.42.3
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1