wav2vec2-gcf-r10

This model is a fine-tuned version of LLL-CREAM/wav2vec2-HAT-0.2K-ALH-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.8386
• Wer: 96.07

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: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH 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
• training_steps: 8000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
16.5882	0.9390	400	3.9077	96.07
7.5377	1.8779	800	3.8877	96.07
7.5493	2.8169	1200	3.8616	96.07
7.5156	3.7559	1600	3.8840	96.07
7.5458	4.6948	2000	3.8063	96.07
7.5305	5.6338	2400	3.9435	96.07
7.7807	6.5728	2800	3.8802	96.07
7.4793	7.5117	3200	3.9080	96.07
7.5548	8.4507	3600	3.8987	96.07
7.5145	9.3897	4000	3.8322	96.07
7.5244	10.3286	4400	3.8388	96.07
7.5244	11.2676	4800	3.8251	96.07
7.5201	12.2066	5200	3.8584	96.07
7.5033	13.1455	5600	3.8279	96.07
7.5042	14.0845	6000	3.8672	96.07
7.5237	15.0235	6400	3.8980	96.07
7.5074	15.9624	6800	3.8657	96.07
7.5174	16.9014	7200	3.8584	96.07
7.5236	17.8404	7600	3.8521	96.07
7.4992	18.7793	8000	3.8386	96.07

Framework versions

• Transformers 5.5.0
• Pytorch 2.4.1+cu124
• Datasets 3.6.0
• Tokenizers 0.22.2