metadata

license: apache-2.0
base_model: facebook/wav2vec2-large-xlsr-53
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: wav2vec2-xlsr-53-ft-btb-ccv-cy
    results: []

wav2vec2-xlsr-53-ft-btb-ccv-cy

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.4198
Wer: 0.3249

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: 0.0003
train_batch_size: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 2600
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0.1046	100	3.5059	1.0
No log	0.2092	200	3.2687	1.0
No log	0.3138	300	2.7369	1.0
No log	0.4184	400	1.3039	0.8545
3.5426	0.5230	500	1.0518	0.7738
3.5426	0.6276	600	0.8396	0.6149
3.5426	0.7322	700	0.7548	0.5687
3.5426	0.8368	800	0.6869	0.5185
3.5426	0.9414	900	0.6501	0.4895
0.7785	1.0460	1000	0.5817	0.4501
0.7785	1.1506	1100	0.5647	0.4288
0.7785	1.2552	1200	0.5424	0.4279
0.7785	1.3598	1300	0.5264	0.4051
0.7785	1.4644	1400	0.5099	0.3977
0.5795	1.5690	1500	0.5058	0.3953
0.5795	1.6736	1600	0.4804	0.3789
0.5795	1.7782	1700	0.4672	0.3698
0.5795	1.8828	1800	0.4605	0.3712
0.5795	1.9874	1900	0.4491	0.3556
0.5057	2.0921	2000	0.4494	0.3453
0.5057	2.1967	2100	0.4432	0.3397
0.5057	2.3013	2200	0.4378	0.3351
0.5057	2.4059	2300	0.4291	0.3310
0.5057	2.5105	2400	0.4279	0.3294
0.3986	2.6151	2500	0.4234	0.3259
0.3986	2.7197	2600	0.4198	0.3249

Framework versions

Transformers 4.40.2
Pytorch 2.3.0+cu121
Datasets 2.19.1
Tokenizers 0.19.1