finetune-large-v2-03102025-exp2

This model is a fine-tuned version of openai/whisper-large-v2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0841
• Wer: 4.0248
• Cer: 1.7761

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: 64
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• 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_steps: 116
• training_steps: 580

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.289	0.9915	58	0.1063	5.8824	2.0452
0.1082	1.9744	116	0.0920	4.6440	1.8299
0.0568	2.9573	174	0.1029	4.3344	1.4532
0.0294	3.9402	232	0.0737	2.7864	1.2379
0.0122	4.9231	290	0.0966	4.0248	1.9914
0.0065	5.9060	348	0.0822	3.4056	1.3455
0.0041	6.8889	406	0.0808	4.3344	1.9914
0.0029	7.8718	464	0.0843	4.0248	1.7761
0.002	8.8547	522	0.0839	4.0248	1.7761
0.003	9.8376	580	0.0841	4.0248	1.7761

Framework versions

• Transformers 4.51.1
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4