speecht5_finetuned_voxpopuli_fr

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

• Loss: 0.4697

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss
0.765	0.23	50	0.6575
0.687	0.47	100	0.6106
0.6423	0.7	150	0.5548
0.5792	0.94	200	0.5300
0.5658	1.17	250	0.5186
0.5558	1.41	300	0.5078
0.5484	1.64	350	0.5029
0.5427	1.87	400	0.4981
0.5349	2.11	450	0.4921
0.524	2.34	500	0.4906
0.5243	2.58	550	0.4857
0.5238	2.81	600	0.4835
0.5104	3.05	650	0.4796
0.516	3.28	700	0.4769
0.5084	3.51	750	0.4763
0.5029	3.75	800	0.4749
0.5015	3.98	850	0.4725
0.5045	4.22	900	0.4716
0.503	4.45	950	0.4706
0.5013	4.69	1000	0.4697

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.2
• Tokenizers 0.13.3