Eimhin03/Prereained_Synthetic_Data_augmentation_fine_tuned

irish-whisper-combined-aug_attempt3

This model is a fine-tuned version of Eimhin03/output_model_Eubookshop_data_base_model_100k on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8665
• Wer: 38.6966

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: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• 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: 1000
• training_steps: 30000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1914	8.8347	2500	0.9888	48.0846
0.0616	17.6684	5000	0.9380	47.5215
0.0211	26.5022	7500	0.8980	42.9244
0.0258	35.3360	10000	0.8616	41.1890
0.0464	44.1698	12500	0.9682	44.8352
0.0380	53.0035	15000	0.9432	43.5429
0.0378	61.8382	17500	0.9383	41.6782
0.0124	70.6720	20000	0.9222	41.5859
0.0242	79.5057	22500	0.9148	40.9582
0.0089	88.3395	25000	0.8964	42.0474
0.0091	97.1733	27500	0.8680	38.6966
0.0132	106.0071	30000	0.8665	38.6966

Framework versions

• Transformers 5.3.0.dev0
• Pytorch 2.9.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.2