metadata

license: apache-2.0
tags:
  - generated_from_trainer
model-index:
  - name: Wav2Vec2_FullDataset
    results: []
datasets:
  - timit-asr/timit_asr
language:
  - en
metrics:
  - wer
base_model:
  - facebook/wav2vec2-base
pipeline_tag: automatic-speech-recognition
library_name: transformers

Wav2Vec2_FullDataset

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

Loss: 0.5144
Wer: 0.3292

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
3.5171	1.0	500	1.8108	1.0014
0.8397	2.01	1000	0.5559	0.5358
0.431	3.01	1500	0.4265	0.4469
0.2931	4.02	2000	0.4034	0.4193
0.2247	5.02	2500	0.4595	0.4076
0.1855	6.02	3000	0.4543	0.3991
0.1497	7.03	3500	0.4894	0.3839
0.1339	8.03	4000	0.4514	0.3836
0.1166	9.04	4500	0.4432	0.3682
0.1063	10.04	5000	0.4781	0.3773
0.0923	11.04	5500	0.4548	0.3699
0.0899	12.05	6000	0.4836	0.3636
0.0802	13.05	6500	0.5117	0.3637
0.0726	14.06	7000	0.4453	0.3653
0.07	15.06	7500	0.4983	0.3581
0.0641	16.06	8000	0.4922	0.3603
0.0561	17.07	8500	0.4947	0.3517
0.0522	18.07	9000	0.5132	0.3513
0.0483	19.08	9500	0.4815	0.3453
0.0419	20.08	10000	0.5556	0.3459
0.0402	21.08	10500	0.5141	0.3428
0.0368	22.09	11000	0.5176	0.3437
0.0322	23.09	11500	0.5326	0.3403
0.0305	24.1	12000	0.5046	0.3366
0.0258	25.1	12500	0.5219	0.3315
0.0254	26.1	13000	0.5166	0.3289
0.0226	27.11	13500	0.5177	0.3311
0.0226	28.11	14000	0.5187	0.3302
0.0209	29.12	14500	0.5144	0.3292

Framework versions

Transformers 4.17.0
Pytorch 2.5.1+cu121
Datasets 1.18.3
Tokenizers 0.20.3