DynamicWav2Vec_TEST_15

This model is a fine-tuned version of joorock12/wav2vec2-large-xlsr-italian on the common_voice_1_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0991
• Wer: 0.1932

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: 0.0001
• 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
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2942	1.67	500	0.0745	0.1919
0.2091	3.34	1000	0.0799	0.1954
0.1895	5.02	1500	0.0818	0.1958
0.1673	6.69	2000	0.0891	0.1987
0.1493	8.36	2500	0.0853	0.2004
0.1389	10.03	3000	0.0896	0.1982
0.135	11.71	3500	0.0954	0.2023
0.1246	13.38	4000	0.0947	0.1979
0.1212	15.05	4500	0.0943	0.1995
0.1123	16.72	5000	0.0962	0.2001
0.1073	18.39	5500	0.1025	0.2047
0.1014	20.07	6000	0.1034	0.1983
0.0967	21.74	6500	0.1006	0.1976
0.0968	23.41	7000	0.0969	0.1966
0.092	25.08	7500	0.0980	0.1950
0.0905	26.76	8000	0.0991	0.1943
0.0882	28.43	8500	0.0991	0.1932

Framework versions

• Transformers 4.34.1
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.14.1