wav2vec2-xls-r-300m-lg-CV-Fleurs-50hrs-v10

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: 0.4833
• Wer: 0.3817
• Cer: 0.0880

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: 0.0003
• train_batch_size: 8
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• 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
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
1.3423	0.9998	2145	0.5793	0.6408	0.1559
0.5736	2.0	4291	0.4512	0.5113	0.1204
0.4672	2.9998	6436	0.4223	0.4699	0.1131
0.4105	4.0	8582	0.3913	0.4571	0.1079
0.3732	4.9998	10727	0.3821	0.4472	0.1045
0.345	6.0	12873	0.3729	0.4245	0.0989
0.3208	6.9998	15018	0.3581	0.4260	0.0967
0.3032	8.0	17164	0.3545	0.3998	0.0929
0.285	8.9998	19309	0.3589	0.4077	0.0939
0.2695	10.0	21455	0.3567	0.4116	0.0953
0.2558	10.9998	23600	0.3546	0.4106	0.0928
0.2456	12.0	25746	0.3678	0.4123	0.0934
0.2337	12.9998	27891	0.3685	0.3897	0.0910
0.2247	14.0	30037	0.3636	0.3928	0.0908
0.2138	14.9998	32182	0.3723	0.3941	0.0904
0.2046	16.0	34328	0.3854	0.3858	0.0893
0.1938	16.9998	36473	0.3770	0.3911	0.0901
0.1852	18.0	38619	0.3773	0.3943	0.0907
0.1786	18.9998	40764	0.3976	0.3750	0.0863
0.1711	20.0	42910	0.3987	0.4006	0.0914
0.162	20.9998	45055	0.4311	0.3829	0.0888
0.157	22.0	47201	0.4232	0.3865	0.0897
0.1523	22.9998	49346	0.4390	0.3854	0.0888
0.1445	24.0	51492	0.4499	0.3795	0.0879
0.1408	24.9998	53637	0.4373	0.3824	0.0890
0.1338	26.0	55783	0.4516	0.3846	0.0896
0.1289	26.9998	57928	0.4771	0.3777	0.0880
0.1259	28.0	60074	0.4822	0.3808	0.0882
0.1212	28.9998	62219	0.4752	0.3835	0.0885
0.1175	30.0	64365	0.4755	0.3771	0.0872
0.1121	30.9998	66510	0.4833	0.3817	0.0880

Framework versions

• Transformers 4.46.3
• Pytorch 2.1.0+cu118
• Datasets 3.1.0
• Tokenizers 0.20.3