MCV_Fleurs_Combined_Irish_ASR_normalised5

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: 0.2735
• Wer: 18.4736

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: 50000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.5034	2.2936	2000	0.3801	38.0854
0.2230	4.5872	4000	0.3193	31.3860
0.1550	6.8807	6000	0.3220	27.7605
0.0800	9.1743	8000	0.3311	26.9833
0.0903	11.4679	10000	0.3224	26.6974
0.0713	13.7615	12000	0.3274	25.2282
0.0389	16.0550	14000	0.3338	24.5061
0.0404	18.3486	16000	0.3261	24.0548
0.0406	20.6422	18000	0.3200	23.7539
0.0279	22.9358	20000	0.3197	23.9194
0.0306	25.2294	22000	0.3227	23.1923
0.0383	27.5229	24000	0.3231	22.4351
0.0294	29.8165	26000	0.3076	22.2144
0.0062	32.1101	28000	0.3107	21.8985
0.0309	34.4037	30000	0.3094	21.8584
0.0158	36.6972	32000	0.3083	20.2588
0.0096	38.9908	34000	0.3049	20.1885
0.0091	41.2844	36000	0.3004	20.5295
0.0047	43.5780	38000	0.2948	19.8877
0.0086	45.8716	40000	0.2913	19.5016
0.0123	48.1651	42000	0.2849	19.7824
0.0073	50.4587	44000	0.2840	19.1856
0.0078	52.7523	46000	0.2827	19.4564
0.0061	55.0459	48000	0.2755	18.7945
0.0070	57.3394	50000	0.2735	18.4736

Framework versions

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