wav2vec2-lv-60-espeak-cv-ft-WCTCv2-phocab-ds-f5

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: 13.2029
• Per: 0.0477

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
1441.4509	0.7194	400	255.1934	1.0
518.5561	1.4388	800	242.9866	1.0
493.4076	2.1583	1200	213.9323	0.9824
404.6164	2.8777	1600	128.5979	0.6589
267.2499	3.5971	2000	62.0884	0.3060
184.9936	4.3165	2400	38.2563	0.1380
147.4093	5.0360	2800	28.6490	0.1120
131.191	5.7554	3200	25.0779	0.0970
120.5205	6.4748	3600	21.8576	0.0753
111.4145	7.1942	4000	20.0957	0.0702
107.2278	7.9137	4400	18.3748	0.0677
101.8162	8.6331	4800	17.2956	0.0635
97.0389	9.3525	5200	18.2921	0.0535
95.541	10.0719	5600	16.9642	0.0552
92.266	10.7914	6000	17.0504	0.0569
88.7644	11.5108	6400	16.7166	0.0560
89.0162	12.2302	6800	15.8472	0.0552
86.7122	12.9496	7200	15.5992	0.0527
86.7122	13.6691	7600	15.6059	0.0527
82.7608	14.3885	8000	15.9941	0.0569
84.1817	15.1079	8400	15.2211	0.0510
82.2823	15.8273	8800	15.6046	0.0493
82.1966	16.5468	9200	15.4974	0.0543
83.0292	17.2662	9600	14.6937	0.0485
79.565	17.9856	10000	14.8158	0.0477
78.7939	18.7050	10400	14.6353	0.0477
77.8892	19.4245	10800	14.7352	0.0502
79.3044	20.1439	11200	14.0838	0.0502
76.4536	20.8633	11600	13.8414	0.0493
74.7527	21.5827	12000	14.0932	0.0477
76.2656	22.3022	12400	14.3761	0.0510
75.5962	23.0216	12800	14.0235	0.0493
76.6569	23.7410	13200	13.8555	0.0477
74.009	24.4604	13600	14.2747	0.0510
75.3227	25.1799	14000	12.8055	0.0477
72.7572	25.8993	14400	13.1407	0.0485
74.7924	26.6187	14800	13.2643	0.0468
72.738	27.3381	15200	12.8235	0.0460
70.9732	28.0576	15600	13.2359	0.0510
72.3229	28.7770	16000	12.7495	0.0535
73.4304	29.4964	16400	13.2029	0.0477

Framework versions

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