not_overfited_vc

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

• Loss: 0.4675

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-06
• train_batch_size: 8
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 64
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.7032	0.4002	100	0.5819
0.6216	0.8004	200	0.5350
0.5847	1.2041	300	0.5165
0.5625	1.6043	400	0.5083
0.5647	2.0080	500	0.5047
0.5529	2.4082	600	0.4992
0.5524	2.8084	700	0.4944
0.54	3.2121	800	0.4924
0.5296	3.6123	900	0.4879
0.5418	4.0160	1000	0.4858
0.5345	4.4162	1100	0.4862
0.5186	4.8164	1200	0.4817
0.5232	5.2201	1300	0.4819
0.5309	5.6203	1400	0.4820
0.5315	6.0240	1500	0.4793
0.5238	6.4242	1600	0.4771
0.5121	6.8244	1700	0.4769
0.5252	7.2281	1800	0.4755
0.5251	7.6283	1900	0.4758
0.5136	8.0320	2000	0.4721
0.5176	8.4322	2100	0.4729
0.5096	8.8324	2200	0.4746
0.5155	9.2361	2300	0.4729
0.5091	9.6363	2400	0.4699
0.5223	10.0400	2500	0.4707
0.5105	10.4402	2600	0.4695
0.5148	10.8404	2700	0.4689
0.5101	11.2441	2800	0.4694
0.5125	11.6443	2900	0.4690
0.5093	12.0480	3000	0.4686
0.5057	12.4482	3100	0.4671
0.5063	12.8484	3200	0.4693
0.5071	13.2521	3300	0.4669
0.5051	13.6523	3400	0.4685
0.5049	14.0560	3500	0.4660
0.5015	14.4562	3600	0.4679
0.5041	14.8564	3700	0.4663
0.5108	15.2601	3800	0.4678
0.5048	15.6603	3900	0.4680
0.508	16.0640	4000	0.4675

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.3.1
• Tokenizers 0.21.0