wav2vec2-xls-r-300m-Fleurs_AMMI_AFRIVOICE_LRSC-ln-20hrs-v2

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3763
• Wer: 0.2252
• Cer: 0.0740

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: 4
• eval_batch_size: 2
• seed: 42
• 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
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
5.6396	1.0	1423	2.8770	1.0	1.0
2.8585	2.0	2846	2.7805	1.0	1.0
2.131	3.0	4269	1.0970	0.7618	0.2105
1.2082	4.0	5692	0.7337	0.5150	0.1455
0.9577	5.0	7115	0.5965	0.4144	0.1220
0.8269	6.0	8538	0.5259	0.3800	0.1142
0.7495	7.0	9961	0.5030	0.3515	0.1058
0.69	8.0	11384	0.4540	0.3472	0.1020
0.6471	9.0	12807	0.4356	0.3291	0.0990
0.6132	10.0	14230	0.4083	0.3299	0.0997
0.5859	11.0	15653	0.4029	0.3063	0.0924
0.5537	12.0	17076	0.4057	0.3201	0.0992
0.5426	13.0	18499	0.3984	0.2917	0.0894
0.5156	14.0	19922	0.3756	0.2850	0.0869
0.5007	15.0	21345	0.3751	0.2812	0.0870
0.485	16.0	22768	0.3957	0.2712	0.0842
0.4757	17.0	24191	0.3705	0.2714	0.0842
0.4596	18.0	25614	0.3626	0.2612	0.0813
0.4478	19.0	27037	0.3639	0.2689	0.0834
0.44	20.0	28460	0.3683	0.2620	0.0816
0.4272	21.0	29883	0.3550	0.2632	0.0846
0.4175	22.0	31306	0.3603	0.2543	0.0804
0.4015	23.0	32729	0.3432	0.2544	0.0803
0.3977	24.0	34152	0.3496	0.2519	0.0792
0.3904	25.0	35575	0.3661	0.2452	0.0773
0.3786	26.0	36998	0.3655	0.2463	0.0782
0.3711	27.0	38421	0.3467	0.2463	0.0790
0.3631	28.0	39844	0.3537	0.2463	0.0783
0.3593	29.0	41267	0.3609	0.2361	0.0756
0.3464	30.0	42690	0.3335	0.2531	0.0820
0.3458	31.0	44113	0.3588	0.2365	0.0750
0.3402	32.0	45536	0.3510	0.2352	0.0751
0.3329	33.0	46959	0.3464	0.2362	0.0762
0.3307	34.0	48382	0.3471	0.2340	0.0762
0.3199	35.0	49805	0.3741	0.2374	0.0765
0.3185	36.0	51228	0.3385	0.2390	0.0767
0.3137	37.0	52651	0.3572	0.2317	0.0743
0.3059	38.0	54074	0.3745	0.2294	0.0734
0.3024	39.0	55497	0.3968	0.2299	0.0741
0.2958	40.0	56920	0.3469	0.2317	0.0756
0.296	41.0	58343	0.3302	0.2495	0.0823
0.2927	42.0	59766	0.3747	0.2261	0.0730
0.2842	43.0	61189	0.3799	0.2216	0.0719
0.2754	44.0	62612	0.3530	0.2602	0.0988
0.2781	45.0	64035	0.3907	0.2237	0.0726
0.2656	46.0	65458	0.3523	0.2397	0.0824
0.2649	47.0	66881	0.3621	0.2289	0.0762
0.2605	48.0	68304	0.3946	0.2259	0.0727
0.2633	49.0	69727	0.3852	0.2233	0.0737
0.2623	50.0	71150	0.3821	0.2247	0.0731
0.2544	51.0	72573	0.3742	0.2226	0.0723
0.2512	52.0	73996	0.3686	0.2229	0.0731
0.2522	53.0	75419	0.3763	0.2252	0.0740

Framework versions

• Transformers 4.46.1
• Pytorch 2.1.0+cu118
• Datasets 3.1.0
• Tokenizers 0.20.1