MCV_Fleurs_Combined_Irish_ASR_normalised4

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

• Loss: 0.2571
• Wer: 18.7193

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
• 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: 20000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4963	2.2936	2000	0.3847	33.9886
0.2000	4.5872	4000	0.3206	27.8758
0.1234	6.8807	6000	0.3079	25.6444
0.0543	9.1743	8000	0.3054	24.4609
0.0443	11.4679	10000	0.2952	22.4401
0.0281	13.7615	12000	0.2865	23.2625
0.0164	16.0550	14000	0.2770	20.8956
0.0110	18.3486	16000	0.2712	20.3941
0.0089	20.6422	18000	0.2621	19.3010
0.0063	22.9358	20000	0.2571	18.7193

Framework versions

• Transformers 5.6.0.dev0
• Pytorch 2.9.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.2