xlsr-aiish-nose

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: 0.3068

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
4.497	1.9704	200	2.6013	1.0
1.5579	3.9409	400	0.1774	0.5513
0.2962	5.9113	600	0.0348	0.3826
0.1312	7.8818	800	0.0160	0.3325
0.1006	9.8522	1000	0.0058	0.3166
0.0806	11.8227	1200	0.0047	0.3117
0.0746	13.7931	1400	0.0014	0.3105
0.0548	15.7635	1600	0.0014	0.3093
0.0468	17.7340	1800	0.0009	0.3093
0.053	19.7044	2000	0.0299	0.3117
0.0516	21.6749	2200	0.0034	0.3105
0.0286	23.6453	2400	0.0003	0.3068
0.0376	25.6158	2600	0.0004	0.3068
0.0293	27.5862	2800	0.0003	0.3068
0.0257	29.5567	3000	0.0002	0.3068
0.0212	31.5271	3200	0.0003	0.3093
0.0249	33.4975	3400	0.0007	0.3068
0.0192	35.4680	3600	0.0002	0.3081
0.0221	37.4384	3800	0.0008	0.3142
0.0181	39.4089	4000	0.0003	0.3105
0.0192	41.3793	4200	0.0010	0.3093
0.0219	43.3498	4400	0.0010	0.3117
0.0139	45.3202	4600	0.0010	0.3105
0.0125	47.2906	4800	0.0001	0.3068
0.0107	49.2611	5000	0.0002	0.3068
0.0119	51.2315	5200	0.0128	0.3240
0.0104	53.2020	5400	0.0001	0.3068
0.0093	55.1724	5600	0.0001	0.3142
0.0085	57.1429	5800	0.0001	0.3068
0.0087	59.1133	6000	0.0001	0.3068
0.0082	61.0837	6200	0.0001	0.3068
0.0079	63.0542	6400	0.0001	0.3068
0.0083	65.0246	6600	0.0001	0.3068
0.0056	66.9951	6800	0.0001	0.3068
0.0099	68.9655	7000	0.0001	0.3068
0.0052	70.9360	7200	0.0000	0.3068
0.0051	72.9064	7400	0.0000	0.3068
0.0061	74.8768	7600	0.0001	0.3068
0.0035	76.8473	7800	0.0000	0.3081
0.0051	78.8177	8000	0.0000	0.3068
0.0037	80.7882	8200	0.0000	0.3068
0.0036	82.7586	8400	0.0000	0.3068
0.0041	84.7291	8600	0.0002	0.3068
0.0031	86.6995	8800	0.0006	0.3068
0.0026	88.6700	9000	0.0000	0.3068
0.0017	90.6404	9200	0.0000	0.3068
0.0023	92.6108	9400	0.0000	0.3068
0.0018	94.5813	9600	0.0000	0.3068
0.002	96.5517	9800	0.0000	0.3068
0.0018	98.5222	10000	0.0000	0.3068

Framework versions

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