speecht5_dhivehi_tts_v8_from_scratch

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

• Loss: 0.7747

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: 128
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 256
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine_with_restarts
• lr_scheduler_warmup_steps: 2000
• training_steps: 100000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.3751	22.9942	2000	1.3367
1.3348	45.9827	4000	1.3422
1.2871	68.9711	6000	1.3045
1.2354	91.9595	8000	1.2275
1.1849	114.9480	10000	1.1799
1.1335	137.9364	12000	1.1210
1.0861	160.9249	14000	1.0787
0.9836	183.9133	16000	1.0411
0.8547	206.9017	18000	1.0122
0.799	229.8902	20000	1.0109
0.7594	252.8786	22000	0.9857
0.7332	275.8671	24000	0.9696
0.7104	298.8555	26000	0.9611
0.6849	321.8439	28000	0.9329
0.6566	344.8324	30000	0.9309
0.634	367.8208	32000	0.8978
0.6138	390.8092	34000	0.8874
0.5951	413.7977	36000	0.8814
0.5785	436.7861	38000	0.8510
0.5617	459.7746	40000	0.8569
0.5484	482.7630	42000	0.8135
0.5349	505.7514	44000	0.8215
0.5247	528.7399	46000	0.8190
0.5134	551.7283	48000	0.8198
0.505	574.7168	50000	0.7792
0.494	597.7052	52000	0.8174
0.4862	620.6936	54000	0.7625
0.4783	643.6821	56000	0.7940
0.4731	666.6705	58000	0.7974
0.4667	689.6590	60000	0.7664
0.4621	712.6474	62000	0.7744
0.4554	735.6358	64000	0.7747

Framework versions

• Transformers 4.48.0.dev0
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0