wav2vec2-Y_speed2

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.9944
• Cer: 38.7688

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: 50
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
43.7135	0.1289	200	5.1032	100.0
5.0439	0.2579	400	4.6765	100.0
4.8774	0.3868	600	4.6557	100.0
4.8641	0.5158	800	4.6426	100.0
4.7995	0.6447	1000	4.6472	100.0
4.7762	0.7737	1200	4.6153	100.0
4.7579	0.9026	1400	4.6154	100.0
4.7082	1.0316	1600	4.6180	100.0
4.6786	1.1605	1800	4.5371	100.0
4.6438	1.2895	2000	4.5289	100.0
4.5663	1.4184	2200	4.4416	100.0
4.503	1.5474	2400	4.3983	99.3421
4.2564	1.6763	2600	4.0853	82.7714
3.7092	1.8053	2800	3.2871	61.8656
3.071	1.9342	3000	2.9127	53.0663
2.704	2.0632	3200	2.6764	49.4302
2.4656	2.1921	3400	2.4448	45.2420
2.2855	2.3211	3600	2.2835	42.6339
2.1728	2.4500	3800	2.2042	42.0876
2.0623	2.5790	4000	2.1021	39.8144
1.9909	2.7079	4200	2.0544	39.5266
1.9129	2.8369	4400	2.0083	38.7453
1.9151	2.9658	4600	1.9944	38.7688

Framework versions

• Transformers 4.44.2
• Pytorch 2.5.0+cu121
• Datasets 3.1.0
• Tokenizers 0.19.1