Wav2vec2_MyST_Train_and_Dev

This model is a fine-tuned version of facebook/wav2vec2-base-960h on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6956
• Wer: 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: 0.0001
• train_batch_size: 12
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 24
• 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: 500
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.8218	0.4356	1000	0.6958	1.0
0.7378	0.8713	2000	0.8468	1.0
0.0	1.3067	3000	nan	1.0
0.0	1.7423	4000	nan	1.0
0.0	2.1777	5000	nan	1.0
0.0	2.6134	6000	nan	1.0
0.0	3.0488	7000	nan	1.0
0.0	3.4844	8000	nan	1.0
0.0	3.9201	9000	nan	1.0
0.0	4.3555	10000	nan	1.0
0.0	4.7911	11000	nan	1.0
0.0	5.2265	12000	nan	1.0
0.0	5.6622	13000	nan	1.0
0.0	6.0976	14000	nan	1.0
0.0	6.5332	15000	nan	1.0
0.0	6.9689	16000	nan	1.0
0.0	7.4043	17000	nan	1.0
0.0	7.8399	18000	nan	1.0
0.0	8.2753	19000	nan	1.0
0.0	8.7110	20000	nan	1.0
0.0	9.1464	21000	nan	1.0
0.0	9.5820	22000	nan	1.0

Framework versions

• Transformers 4.56.2
• Pytorch 2.8.0.dev20250319+cu128
• Datasets 4.1.1
• Tokenizers 0.22.1