whisper-tiny-fi

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: 0.7638
• Wer: 309.7839

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
1.986	0.3690	100	1.5979	83.1116
0.755	0.7380	200	0.7632	82.2813
0.57	1.1070	300	0.7001	75.4128
0.517	1.4760	400	0.6558	76.1110
0.4948	1.8450	500	0.6328	71.7426
0.3598	2.2140	600	0.6191	69.7519
0.3708	2.5830	700	0.6093	71.5067
0.3379	2.9520	800	0.5944	70.6010
0.2184	3.3210	900	0.5993	69.8085
0.2335	3.6900	1000	0.5836	69.1197
0.1763	4.0590	1100	0.5925	69.6292
0.1648	4.4280	1200	0.5940	72.7805
0.1471	4.7970	1300	0.5947	74.0542
0.0922	5.1661	1400	0.6138	72.4974
0.0989	5.5351	1500	0.6071	73.5541
0.095	5.9041	1600	0.6121	75.1392
0.0554	6.2731	1700	0.6237	76.0732
0.0606	6.6421	1800	0.6240	79.8000
0.0544	7.0111	1900	0.6418	83.9419
0.0372	7.3801	2000	0.6391	91.3105
0.0414	7.7491	2100	0.6471	81.3850
0.0223	8.1181	2200	0.6521	104.4249
0.0256	8.4871	2300	0.6587	104.8684
0.0233	8.8561	2400	0.6669	119.1056
0.0159	9.2251	2500	0.6907	107.2271
0.0162	9.5941	2600	0.6879	140.2585
0.0156	9.9631	2700	0.6933	185.6024
0.01	10.3321	2800	0.6958	259.4584
0.0099	10.7011	2900	0.7037	205.2363
0.0074	11.0701	3000	0.7080	246.1836
0.0074	11.4391	3100	0.7141	240.3906
0.0074	11.8081	3200	0.7159	196.5185
0.0053	12.1771	3300	0.7246	216.1242
0.0057	12.5461	3400	0.7310	215.3033
0.0056	12.9151	3500	0.7343	232.3521
0.0044	13.2841	3600	0.7374	234.0976
0.0047	13.6531	3700	0.7420	248.5989
0.0046	14.0221	3800	0.7482	245.2684
0.0041	14.3911	3900	0.7480	270.2236
0.0038	14.7601	4000	0.7481	294.0466
0.0037	15.1292	4100	0.7547	263.7513
0.0037	15.4982	4200	0.7551	280.0359
0.0035	15.8672	4300	0.7568	270.1198
0.0032	16.2362	4400	0.7574	286.9327
0.0032	16.6052	4500	0.7611	286.9516
0.0035	16.9742	4600	0.7618	309.7368
0.0032	17.3432	4700	0.7632	298.6508
0.0031	17.7122	4800	0.7632	304.3778
0.0029	18.0812	4900	0.7637	304.8306
0.003	18.4502	5000	0.7638	309.7839

Framework versions

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