speecht5_finetuned_voxpopuli_de

This model is a fine-tuned version of microsoft/speecht5_tts on the facebook/voxpopuli dataset. It achieves the following results on the evaluation set:

• Loss: 0.4565

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• 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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
5.1518	0.5319	250	0.5899
4.4208	1.0638	500	0.5002
4.2216	1.5957	750	0.4824
4.1452	2.1277	1000	0.4751
4.1011	2.6596	1250	0.4713
4.0454	3.1915	1500	0.4669
4.0224	3.7234	1750	0.4646
4.0026	4.2553	2000	0.4626
4.028	4.7872	2250	0.4604
3.9629	5.3191	2500	0.4595
3.9708	5.8511	2750	0.4593
3.9927	6.3830	3000	0.4577
3.9592	6.9149	3250	0.4573
3.952	7.4468	3500	0.4579
3.9651	7.9787	3750	0.4567
3.9483	8.5106	4000	0.4565

Framework versions

• Transformers 4.46.1
• Pytorch 2.3.1
• Datasets 3.0.2
• Tokenizers 0.20.1