wav2vec2-lv-60-espeak-cv-ft-WCTCv2-phocab-ds-f4

This model is a fine-tuned version of facebook/wav2vec2-lv-60-espeak-cv-ft on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 16.0989
• Per: 0.0479

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: 3e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Per
1457.8689	0.7194	400	273.4371	1.0
529.9854	1.4388	800	254.2645	1.0
489.993	2.1583	1200	220.8327	0.9877
361.9673	2.8777	1600	106.9450	0.4706
219.9257	3.5971	2000	47.5095	0.1649
160.2832	4.3165	2400	34.7040	0.1133
134.796	5.0360	2800	29.5108	0.0871
121.0879	5.7554	3200	27.0536	0.0690
114.8673	6.4748	3600	25.1611	0.0741
107.6301	7.1942	4000	23.5403	0.0668
104.0335	7.9137	4400	22.7524	0.0603
97.7448	8.6331	4800	21.7806	0.0683
97.5668	9.3525	5200	20.8315	0.0545
94.3444	10.0719	5600	19.8112	0.0581
92.0796	10.7914	6000	19.2101	0.0552
88.6588	11.5108	6400	18.4504	0.0516
89.2751	12.2302	6800	17.5154	0.0574
85.6948	12.9496	7200	17.4255	0.0472
86.8838	13.6691	7600	17.6383	0.0559
81.2887	14.3885	8000	17.3350	0.0523
81.9403	15.1079	8400	16.5314	0.0458
80.5417	15.8273	8800	17.0623	0.0537
82.2371	16.5468	9200	16.2990	0.0501
78.3855	17.2662	9600	17.1837	0.0479
76.8071	17.9856	10000	15.9834	0.0494
76.1785	18.7050	10400	15.4747	0.0465
76.7258	19.4245	10800	15.5382	0.0501
75.0845	20.1439	11200	16.5199	0.0472
75.8446	20.8633	11600	16.3991	0.0508
75.8891	21.5827	12000	15.7068	0.0443
76.8147	22.3022	12400	16.7615	0.0458
72.8343	23.0216	12800	16.0542	0.0472
74.2895	23.7410	13200	16.4808	0.0479
76.0104	24.4604	13600	17.0120	0.0458
74.7333	25.1799	14000	16.0332	0.0472
73.3391	25.8993	14400	14.8035	0.0428
74.1157	26.6187	14800	15.6057	0.0465
72.7293	27.3381	15200	16.0070	0.0487
71.8061	28.0576	15600	16.1621	0.0508
70.5643	28.7770	16000	15.8943	0.0501
71.6839	29.4964	16400	16.0989	0.0479

Framework versions

• Transformers 4.57.6
• Pytorch 2.9.1+cu128
• Datasets 4.5.0
• Tokenizers 0.22.2