xlsr-aiish-clp

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: 0.3154

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
5.176	2.4845	200	2.7099	1.0
2.5595	4.9689	400	1.7974	1.0
1.1648	7.4534	600	0.1544	0.6039
0.3992	9.9379	800	0.0722	0.5465
0.2507	12.4224	1000	0.0451	0.4670
0.176	14.9068	1200	0.0316	0.3729
0.1324	17.3913	1400	0.0155	0.3655
0.1193	19.8758	1600	0.0075	0.3313
0.0897	22.3602	1800	0.0201	0.3313
0.0825	24.8447	2000	0.0035	0.3117
0.054	27.3292	2200	0.0058	0.3154
0.0579	29.8137	2400	0.0013	0.3081
0.0529	32.2981	2600	0.0028	0.3093
0.0418	34.7826	2800	0.0043	0.3081
0.0402	37.2671	3000	0.0084	0.3117
0.0366	39.7516	3200	0.0006	0.3093
0.0326	42.2360	3400	0.0031	0.3117
0.0266	44.7205	3600	0.0029	0.3166
0.0268	47.2050	3800	0.0009	0.3081
0.024	49.6894	4000	0.0014	0.3093
0.0185	52.1739	4200	0.0001	0.3081
0.0277	54.6584	4400	0.0011	0.3068
0.0179	57.1429	4600	0.0026	0.3093
0.0161	59.6273	4800	0.0086	0.3142
0.0177	62.1118	5000	0.0004	0.3130
0.0184	64.5963	5200	0.0009	0.3130
0.0131	67.0807	5400	0.0022	0.3166
0.017	69.5652	5600	0.0026	0.3105
0.0083	72.0497	5800	0.0023	0.3081
0.0082	74.5342	6000	0.0001	0.3093
0.0068	77.0186	6200	0.0001	0.3142
0.0072	79.5031	6400	0.0002	0.3142
0.0097	81.9876	6600	0.0001	0.3142
0.0087	84.4720	6800	0.0001	0.3154
0.0065	86.9565	7000	0.0000	0.3154
0.0072	89.4410	7200	0.0000	0.3154
0.0065	91.9255	7400	0.0001	0.3154
0.0044	94.4099	7600	0.0001	0.3154
0.0053	96.8944	7800	0.0000	0.3154
0.005	99.3789	8000	0.0000	0.3154

Framework versions

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