metadata

license: apache-2.0
tags:
  - generated_from_trainer
model-index:
  - name: XLRS_FullDataset
    results: []
datasets:
  - timit-asr/timit_asr
language:
  - en
base_model:
  - facebook/wav2vec2-base
pipeline_tag: automatic-speech-recognition
metrics:
  - wer
library_name: transformers

XLRS_FullDataset

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: 0.3057
Wer: 0.2697

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: 0.0001
train_batch_size: 8
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 1000
num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.5696	1.0	500	3.1546	1.0
2.491	2.01	1000	0.8309	0.7872
0.7519	3.01	1500	0.3648	0.4364
0.4704	4.02	2000	0.2998	0.3758
0.3385	5.02	2500	0.2639	0.3439
0.2837	6.02	3000	0.2604	0.3309
0.2233	7.03	3500	0.2734	0.3143
0.1997	8.03	4000	0.2676	0.3121
0.1717	9.04	4500	0.2489	0.2941
0.1558	10.04	5000	0.2777	0.2969
0.1497	11.04	5500	0.2693	0.2890
0.1326	12.05	6000	0.2844	0.2921
0.118	13.05	6500	0.2818	0.2969
0.119	14.06	7000	0.2798	0.2854
0.0991	15.06	7500	0.2765	0.2858
0.108	16.06	8000	0.2904	0.2794
0.0935	17.07	8500	0.2846	0.2773
0.0857	18.07	9000	0.3120	0.2812
0.0928	19.08	9500	0.3073	0.2820
0.0832	20.08	10000	0.2981	0.2808
0.0768	21.08	10500	0.3065	0.2807
0.0768	22.09	11000	0.2960	0.2766
0.0754	23.09	11500	0.3007	0.2783
0.063	24.1	12000	0.2918	0.2739
0.0614	25.1	12500	0.3144	0.2748
0.0628	26.1	13000	0.3074	0.2713
0.0595	27.11	13500	0.3103	0.2695
0.0616	28.11	14000	0.3108	0.2697
0.0587	29.12	14500	0.3057	0.2697

Framework versions

Transformers 4.17.0
Pytorch 2.5.1+cu121
Datasets 1.18.3
Tokenizers 0.20.3