w2v-bert-2.0-lmk_man_align_updated

This model is a fine-tuned version of facebook/w2v-bert-2.0 on the audiofolder dataset. It achieves the following results on the evaluation set:

• Loss: nan
• Wer: 1.0
• Cer: 1.0

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: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 300
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
3.6561	6.25	100	3.1439	1.0	0.9246
5.5678	12.5	200	4.0590	1.1777	0.7807
5.583	18.75	300	4.0602	1.1533	0.7807
5.2971	25.0	400	4.0607	1.1498	0.7845
5.6771	31.25	500	4.0620	1.1568	0.7791
0.0	37.5	600	nan	1.0	1.0
0.0	43.75	700	nan	1.0	1.0
0.0	50.0	800	nan	1.0	1.0
0.0	56.25	900	nan	1.0	1.0
0.0	62.5	1000	nan	1.0	1.0
0.0	68.75	1100	nan	1.0	1.0
0.0	75.0	1200	nan	1.0	1.0
0.0	81.25	1300	nan	1.0	1.0
0.0	87.5	1400	nan	1.0	1.0
0.0	93.75	1500	nan	1.0	1.0
0.0	100.0	1600	nan	1.0	1.0

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu128
• Datasets 3.0.0
• Tokenizers 0.22.1