speecht5_finetuned_Bengali_shabdobhedii

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: 0.5902

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: 4
• eval_batch_size: 4
• 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: 100
• training_steps: 1500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.7972	0.3388	100	0.7451
0.7235	0.6777	200	0.6852
0.6903	1.0165	300	0.6570
0.6906	1.3554	400	0.6403
0.676	1.6942	500	0.6299
0.6851	2.0330	600	0.6257
0.6583	2.3719	700	0.6170
0.654	2.7107	800	0.6163
0.6419	3.0496	900	0.6112
0.6344	3.3884	1000	0.6055
0.626	3.7272	1100	0.6002
0.6203	4.0661	1200	0.5949
0.6253	4.4049	1300	0.5938
0.6127	4.7438	1400	0.5894
0.6104	5.0826	1500	0.5902

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.19.1