mms-1b-all-lin-Fleurs_AMMI_AFRIVOICE_LRSC-50hrs-v1

This model is a fine-tuned version of facebook/mms-1b-all on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2703
• Wer: 0.1898
• Cer: 0.0562

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 4
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
2.5343	1.0	1424	0.3279	0.2353	0.0681
0.4295	2.0	2848	0.3095	0.2265	0.0661
0.4103	3.0	4272	0.2973	0.2279	0.0662
0.3942	4.0	5696	0.2880	0.2209	0.0649
0.384	5.0	7120	0.2838	0.2213	0.0649
0.3751	6.0	8544	0.2758	0.2170	0.0639
0.369	7.0	9968	0.2763	0.2205	0.0642
0.3631	8.0	11392	0.2687	0.2149	0.0629
0.359	9.0	12816	0.2662	0.2130	0.0624
0.3521	10.0	14240	0.2726	0.2076	0.0617
0.3496	11.0	15664	0.2629	0.2078	0.0617
0.3455	12.0	17088	0.2574	0.2109	0.0618
0.3422	13.0	18512	0.2619	0.2051	0.0606
0.3385	14.0	19936	0.2542	0.2092	0.0610
0.3333	15.0	21360	0.2530	0.2050	0.0601
0.3312	16.0	22784	0.2550	0.2018	0.0591
0.3278	17.0	24208	0.2551	0.1983	0.0587
0.3265	18.0	25632	0.2653	0.1974	0.0586
0.3222	19.0	27056	0.2542	0.1990	0.0589
0.3171	20.0	28480	0.2566	0.1966	0.0587
0.3155	21.0	29904	0.2533	0.1984	0.0588
0.3122	22.0	31328	0.2577	0.1961	0.0582
0.3105	23.0	32752	0.2556	0.1950	0.0577
0.3053	24.0	34176	0.2526	0.1939	0.0576
0.3068	25.0	35600	0.2563	0.1932	0.0576
0.3017	26.0	37024	0.2667	0.1908	0.0568
0.2984	27.0	38448	0.2551	0.1932	0.0573
0.2956	28.0	39872	0.2591	0.1954	0.0582
0.2921	29.0	41296	0.2668	0.1918	0.0572
0.2876	30.0	42720	0.2598	0.1904	0.0571
0.2872	31.0	44144	0.2637	0.1938	0.0579
0.2825	32.0	45568	0.2649	0.1932	0.0575
0.2808	33.0	46992	0.2527	0.2134	0.0671
0.2773	34.0	48416	0.2555	0.1963	0.0591
0.2752	35.0	49840	0.2569	0.1936	0.0584
0.2719	36.0	51264	0.2703	0.1898	0.0562

Framework versions

• Transformers 4.48.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0