waxal-whisper_small_multi_s42

This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4327
• Wer: 0.7159
• Cer: 0.3848
• Score: 0.5504

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• 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_ratio: 0.1
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer	Score
0.5996	0.5496	1000	0.5983	0.8222	0.4107	0.6165
0.4298	1.0989	2000	0.4846	0.5734	0.2759	0.4246
0.3923	1.6485	3000	0.4546	0.5380	0.2517	0.3949
0.2835	2.1979	4000	0.4413	0.4935	0.2347	0.3641
0.3116	2.7475	5000	0.4295	0.6399	0.3235	0.4817
0.2426	3.2968	6000	0.4358	0.6338	0.3496	0.4917
0.2553	3.8464	7000	0.4327	0.7159	0.3848	0.5504

Framework versions

• Transformers 4.57.6
• Pytorch 2.10.0+cu128
• Datasets 3.6.0
• Tokenizers 0.22.2