wav2vec2-btb-cv-ft-btb-cy-cand

This model is a fine-tuned version of DewiBrynJones/wav2vec2-xlsr-53-ft-btb-cv-cy on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: inf
• Wer: 0.3402

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: 4
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• training_steps: 10000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0.0215	200	inf	0.5592
No log	0.0429	400	inf	0.4289
2.1964	0.0644	600	inf	0.4374
2.1964	0.0858	800	inf	0.4944
0.8327	0.1073	1000	inf	0.5150
0.8327	0.1287	1200	inf	0.5634
0.8327	0.1502	1400	inf	0.5355
0.91	0.1716	1600	inf	0.5152
0.91	0.1931	1800	inf	0.5595
0.8721	0.2145	2000	inf	0.5057
0.8721	0.2360	2200	inf	0.5041
0.8721	0.2574	2400	inf	0.5146
0.8218	0.2789	2600	inf	0.5018
0.8218	0.3003	2800	inf	0.5091
0.8469	0.3218	3000	inf	0.5037
0.8469	0.3432	3200	inf	0.4703
0.8469	0.3647	3400	inf	0.4795
0.8142	0.3861	3600	inf	0.4714
0.8142	0.4076	3800	inf	0.4554
0.8085	0.4290	4000	inf	0.4506
0.8085	0.4505	4200	inf	0.4458
0.8085	0.4720	4400	inf	0.4367
0.7802	0.4934	4600	inf	0.4401
0.7802	0.5149	4800	inf	0.4334
0.7493	0.5363	5000	inf	0.4224
0.7493	0.5578	5200	inf	0.4328
0.7493	0.5792	5400	inf	0.4176
0.7668	0.6007	5600	inf	0.4183
0.7668	0.6221	5800	inf	0.4030
0.6999	0.6436	6000	inf	0.4125
0.6999	0.6650	6200	inf	0.4076
0.6999	0.6865	6400	inf	0.3917
0.6918	0.7079	6600	inf	0.4004
0.6918	0.7294	6800	inf	0.3865
0.6888	0.7508	7000	inf	0.3785
0.6888	0.7723	7200	inf	0.3824
0.6888	0.7937	7400	inf	0.3743
0.646	0.8152	7600	inf	0.3673
0.646	0.8366	7800	inf	0.3667
0.6324	0.8581	8000	inf	0.3662
0.6324	0.8795	8200	inf	0.3601
0.6324	0.9010	8400	inf	0.3535
0.6221	0.9224	8600	inf	0.3526
0.6221	0.9439	8800	inf	0.3487
0.6215	0.9654	9000	inf	0.3481
0.6215	0.9868	9200	inf	0.3447
0.6215	1.0083	9400	inf	0.3410
0.5603	1.0297	9600	inf	0.3405
0.5603	1.0512	9800	inf	0.3412
0.5284	1.0726	10000	inf	0.3402

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1