wav2vec2-lv-60-espeak-cv-ft-WCTC-phocab-ds-f9

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: 10.3456
• Per: 0.0464

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
1285.5111	0.7194	400	255.6897	1.0
491.2075	1.4388	800	245.1350	1.0
467.6034	2.1583	1200	219.8392	1.0
327.2771	2.8777	1600	69.2714	0.2639
177.5675	3.5971	2000	38.9889	0.1661
134.5549	4.3165	2400	29.5711	0.0987
115.3988	5.0360	2800	24.2336	0.0809
103.2192	5.7554	3200	21.1353	0.0717
94.2958	6.4748	3600	18.8294	0.0641
90.8236	7.1942	4000	17.4744	0.0632
85.971	7.9137	4400	17.2274	0.0641
81.7601	8.6331	4800	16.3193	0.0599
78.9497	9.3525	5200	16.2324	0.0658
77.6978	10.0719	5600	14.9711	0.0632
74.688	10.7914	6000	14.8796	0.0582
73.3821	11.5108	6400	14.4326	0.0565
74.0314	12.2302	6800	14.1802	0.0582
69.3884	12.9496	7200	13.0210	0.0548
68.7881	13.6691	7600	13.2149	0.0523
66.1921	14.3885	8000	12.9118	0.0582
66.9728	15.1079	8400	12.7982	0.0548
65.4125	15.8273	8800	12.3509	0.0540
66.1715	16.5468	9200	12.3883	0.0540
63.7279	17.2662	9600	12.1845	0.0540
63.1032	17.9856	10000	11.8108	0.0514
62.3123	18.7050	10400	11.0995	0.0497
61.9874	19.4245	10800	11.3445	0.0506
61.3819	20.1439	11200	11.0286	0.0489
59.9633	20.8633	11600	11.1077	0.0497
60.6667	21.5827	12000	11.3269	0.0489
60.6234	22.3022	12400	11.2123	0.0506
58.682	23.0216	12800	11.1844	0.0506
59.2748	23.7410	13200	11.0940	0.0497
59.9911	24.4604	13600	10.6326	0.0506
56.9586	25.1799	14000	10.6550	0.0481
56.9045	25.8993	14400	10.5189	0.0481
56.0553	26.6187	14800	10.5660	0.0514
57.8147	27.3381	15200	10.2959	0.0472
56.5455	28.0576	15600	10.2758	0.0472
56.0882	28.7770	16000	10.3098	0.0481
57.0629	29.4964	16400	10.3456	0.0464

Framework versions

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