wav2vec2-lv-60-espeak-cv-ft-WCTCv2-phocab-ds-f1

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: 15.7278
• Per: 0.0600

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
1406.667	0.7194	400	271.9598	1.0
524.9791	1.4388	800	253.3634	1.0
488.9751	2.1583	1200	218.0370	0.9784
358.9653	2.8777	1600	104.8389	0.5023
222.6761	3.5971	2000	53.3614	0.1846
168.5138	4.3165	2400	39.5178	0.1228
141.3749	5.0360	2800	32.3725	0.1097
125.7726	5.7554	3200	29.8872	0.0947
117.4302	6.4748	3600	26.5132	0.0834
108.8292	7.1942	4000	23.6709	0.0862
107.8627	7.9137	4400	24.1837	0.0834
101.0746	8.6331	4800	21.8784	0.0731
96.8174	9.3525	5200	21.7872	0.0675
96.3013	10.0719	5600	19.5140	0.0665
93.3701	10.7914	6000	19.2951	0.0694
88.3764	11.5108	6400	18.9131	0.0637
91.7877	12.2302	6800	19.9927	0.0759
88.1197	12.9496	7200	19.6466	0.0694
87.061	13.6691	7600	19.2956	0.0712
86.8002	14.3885	8000	19.6747	0.0703
82.0821	15.1079	8400	19.6085	0.0703
82.8969	15.8273	8800	18.5813	0.0675
82.5427	16.5468	9200	17.6526	0.0656
81.9866	17.2662	9600	17.6926	0.0637
81.0838	17.9856	10000	17.0978	0.0628
79.0553	18.7050	10400	18.3001	0.0647
78.4576	19.4245	10800	17.6350	0.0600
78.7845	20.1439	11200	18.5353	0.0609
78.9591	20.8633	11600	16.2576	0.0637
78.736	21.5827	12000	16.6079	0.0609
76.6259	22.3022	12400	17.6662	0.0572
77.002	23.0216	12800	18.0811	0.0628
75.8286	23.7410	13200	15.8958	0.0572
74.4543	24.4604	13600	16.4727	0.0600
75.5397	25.1799	14000	17.5707	0.0525
74.4407	25.8993	14400	15.8212	0.0600
72.5929	26.6187	14800	15.7854	0.0572
74.1899	27.3381	15200	15.6943	0.0637
73.9154	28.0576	15600	16.0071	0.0675
73.497	28.7770	16000	15.8887	0.0600
72.7745	29.4964	16400	15.7278	0.0600

Framework versions

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