speecht5_soome-V2

This model is a fine-tuned version of catsOfpeople/speecht5_finetuned_emirhan_soomea on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2695

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: 16
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• training_steps: 3500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.9648	4.5198	100	0.4308
0.4495	9.0395	200	0.3583
0.384	13.5593	300	0.3418
0.3637	18.0791	400	0.3177
0.3443	22.5989	500	0.3119
0.3366	27.1186	600	0.3099
0.3328	31.6384	700	0.3222
0.3238	36.1582	800	0.3091
0.3196	40.6780	900	0.2960
0.3156	45.1977	1000	0.2977
0.3123	49.7175	1100	0.2960
0.3107	54.2373	1200	0.2904
0.3029	58.7571	1300	0.2891
0.2978	63.2768	1400	0.2904
0.3012	67.7966	1500	0.2855
0.2977	72.3164	1600	0.2863
0.2915	76.8362	1700	0.2855
0.2935	81.3559	1800	0.2853
0.2877	85.8757	1900	0.2794
0.2839	90.3955	2000	0.2820
0.2847	94.9153	2100	0.2781
0.2831	99.4350	2200	0.2799
0.283	103.9548	2300	0.2811
0.2792	108.4746	2400	0.2774
0.2788	112.9944	2500	0.2813
0.2793	117.5141	2600	0.2755
0.2746	122.0339	2700	0.2769
0.2735	126.5537	2800	0.2729
0.2728	131.0734	2900	0.2764
0.2735	135.5932	3000	0.2751
0.2726	140.1130	3100	0.2754
0.2691	144.6328	3200	0.2707
0.2711	149.1525	3300	0.2717
0.2679	153.6723	3400	0.2724
0.2665	158.1921	3500	0.2695

Framework versions

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