speecht5_finetuned_emirhan_tr

This model is a fine-tuned version of MohsenABG/speecht5_finetuned_emirhan_tr on the common_voice_13_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.5274

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: 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: 100
• training_steps: 3000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.6208	0.3972	100	0.6033
0.6239	0.7944	200	0.5985
0.616	1.1917	300	0.5867
0.6057	1.5889	400	0.5781
0.5971	1.9861	500	0.5718
0.5975	2.3833	600	0.5701
0.5979	2.7805	700	0.5746
0.5913	3.1778	800	0.5651
0.5943	3.5750	900	0.5609
0.5924	3.9722	1000	0.5610
0.5879	4.3694	1100	0.5578
0.5768	4.7666	1200	0.5505
0.5816	5.1639	1300	0.5511
0.5779	5.5611	1400	0.5518
0.5773	5.9583	1500	0.5456
0.5729	6.3555	1600	0.5479
0.5656	6.7527	1700	0.5459
0.5724	7.1500	1800	0.5402
0.5769	7.5472	1900	0.5366
0.5622	7.9444	2000	0.5389
0.5569	8.3416	2100	0.5390
0.5626	8.7388	2200	0.5358
0.5602	9.1360	2300	0.5372
0.5533	9.5333	2400	0.5326
0.5515	9.9305	2500	0.5325
0.5495	10.3277	2600	0.5328
0.5456	10.7249	2700	0.5300
0.5413	11.1221	2800	0.5312
0.5468	11.5194	2900	0.5290
0.5386	11.9166	3000	0.5274

Framework versions

• Transformers 4.44.2
• Pytorch 2.3.0+cu118
• Datasets 3.0.0
• Tokenizers 0.19.1