wav2vec2-lv-60-espeak-cv-ft-WCTCv2-phocab-ds-f3

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: 20.7779
• Per: 0.0412

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
1333.0395	0.7194	400	262.2852	1.0
516.1456	1.4388	800	248.1824	1.0
469.3	2.1583	1200	159.4321	0.7994
267.9056	2.8777	1600	54.5876	0.2149
178.1615	3.5971	2000	35.7229	0.1158
149.3252	4.3165	2400	28.5338	0.0761
133.8608	5.0360	2800	24.9404	0.0650
117.9745	5.7554	3200	22.5180	0.0595
111.0293	6.4748	3600	20.7891	0.0595
108.2733	7.1942	4000	20.1303	0.0603
105.1052	7.9137	4400	19.6570	0.0523
98.2039	8.6331	4800	19.3903	0.0539
97.875	9.3525	5200	17.7390	0.0492
93.8711	10.0719	5600	18.8366	0.0523
91.0766	10.7914	6000	17.2691	0.0523
90.6564	11.5108	6400	18.1785	0.0476
91.2836	12.2302	6800	18.4800	0.0523
86.5649	12.9496	7200	17.7615	0.0468
87.641	13.6691	7600	19.4869	0.0484
88.2231	14.3885	8000	19.0391	0.0484
82.9629	15.1079	8400	17.5536	0.0428
83.7759	15.8273	8800	18.1470	0.0468
82.7616	16.5468	9200	18.9932	0.0452
81.0433	17.2662	9600	19.6497	0.0436
83.0983	17.9856	10000	19.2848	0.0460
80.8659	18.7050	10400	19.6904	0.0428
78.1035	19.4245	10800	19.9923	0.0444
79.3873	20.1439	11200	19.6049	0.0452
78.5869	20.8633	11600	20.1387	0.0484
77.8713	21.5827	12000	19.7639	0.0420
77.5057	22.3022	12400	19.8439	0.0484
77.6414	23.0216	12800	20.2742	0.0428
77.3338	23.7410	13200	17.7020	0.0420
78.1314	24.4604	13600	18.7909	0.0412
73.9889	25.1799	14000	20.1497	0.0404
75.1564	25.8993	14400	21.3150	0.0412
75.9133	26.6187	14800	21.4130	0.0428
73.7971	27.3381	15200	20.4961	0.0389
74.2196	28.0576	15600	20.4902	0.0397
73.7867	28.7770	16000	21.3467	0.0412
74.7683	29.4964	16400	20.7779	0.0412

Framework versions

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