xlsr-nomimosev-aiish

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0000
• Wer: 0.0

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: 0.0004
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 132
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.8467	1.2945	200	2.8108	1.0
2.6337	2.5890	400	2.0411	1.0
1.2205	3.8835	600	0.1323	0.4010
0.3704	5.1780	800	0.0457	0.1907
0.2019	6.4725	1000	0.0216	0.1748
0.158	7.7670	1200	0.0050	0.1027
0.1039	9.0615	1400	0.0018	0.0134
0.0988	10.3560	1600	0.0116	0.0159
0.0758	11.6505	1800	0.0061	0.0110
0.0681	12.9450	2000	0.0068	0.0208
0.0549	14.2395	2200	0.0022	0.0232
0.0622	15.5340	2400	0.0005	0.0355
0.0507	16.8285	2600	0.0018	0.0061
0.0442	18.1230	2800	0.0005	0.0012
0.044	19.4175	3000	0.0032	0.0037
0.0428	20.7120	3200	0.0025	0.0098
0.04	22.0065	3400	0.0253	0.0660
0.0385	23.3010	3600	0.0008	0.0098
0.0333	24.5955	3800	0.0003	0.0
0.0289	25.8900	4000	0.0002	0.0
0.0225	27.1845	4200	0.0031	0.0587
0.0275	28.4790	4400	0.0006	0.0220
0.0275	29.7735	4600	0.0005	0.0086
0.0295	31.0680	4800	0.0003	0.0073
0.0281	32.3625	5000	0.0002	0.0061
0.0254	33.6570	5200	0.0005	0.0269
0.0229	34.9515	5400	0.0022	0.0330
0.0203	36.2460	5600	0.0009	0.0147
0.0224	37.5405	5800	0.0001	0.0
0.0176	38.8350	6000	0.0001	0.0086
0.0204	40.1294	6200	0.0004	0.0073
0.0172	41.4239	6400	0.0004	0.0037
0.0157	42.7184	6600	0.0001	0.0
0.0157	44.0129	6800	0.0001	0.0
0.0146	45.3074	7000	0.0001	0.0012
0.0105	46.6019	7200	0.0001	0.0012
0.0122	47.8964	7400	0.0001	0.0
0.014	49.1909	7600	0.0004	0.0012
0.0187	50.4854	7800	0.0001	0.0024
0.0105	51.7799	8000	0.0004	0.0024
0.0094	53.0744	8200	0.0059	0.0037
0.0082	54.3689	8400	0.0000	0.0
0.0106	55.6634	8600	0.0077	0.0159
0.0082	56.9579	8800	0.0000	0.0049
0.0085	58.2524	9000	0.0000	0.0134
0.0054	59.5469	9200	0.0000	0.0049
0.0077	60.8414	9400	0.0000	0.0122
0.0098	62.1359	9600	0.0000	0.0061
0.0094	63.4304	9800	0.0000	0.0049
0.0069	64.7249	10000	0.0001	0.0012
0.0081	66.0194	10200	0.0000	0.0
0.006	67.3139	10400	0.0001	0.0
0.0061	68.6084	10600	0.0004	0.0024
0.005	69.9029	10800	0.0000	0.0
0.0055	71.1974	11000	0.0010	0.0037
0.0053	72.4919	11200	0.0001	0.0012
0.0057	73.7864	11400	0.0000	0.0061
0.0064	75.0809	11600	0.0001	0.0012
0.0068	76.3754	11800	0.0000	0.0012
0.0055	77.6699	12000	0.0019	0.0134
0.0068	78.9644	12200	0.0008	0.0073
0.0045	80.2589	12400	0.0000	0.0061
0.0032	81.5534	12600	0.0000	0.0073
0.0038	82.8479	12800	0.0000	0.0
0.0028	84.1424	13000	0.0000	0.0037
0.004	85.4369	13200	0.0000	0.0
0.0027	86.7314	13400	0.0000	0.0
0.0019	88.0259	13600	0.0000	0.0
0.0019	89.3204	13800	0.0000	0.0
0.0019	90.6149	14000	0.0000	0.0
0.0016	91.9094	14200	0.0000	0.0
0.0016	93.2039	14400	0.0000	0.0012
0.0015	94.4984	14600	0.0000	0.0
0.0013	95.7929	14800	0.0000	0.0
0.0023	97.0874	15000	0.0000	0.0
0.0017	98.3819	15200	0.0000	0.0
0.0018	99.6764	15400	0.0000	0.0

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.4.0
• Datasets 2.21.0
• Tokenizers 0.19.1