metadata

license: apache-2.0
base_model: zainulhakim/240615-wav2vec2-ASR-Arabs
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: 240624-wav2vec2-ASR-Arab
    results: []

240624-wav2vec2-ASR-Arab

This model is a fine-tuned version of zainulhakim/240615-wav2vec2-ASR-Arabs on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 1.3847
Wer: 0.9798

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: 5
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: 500
num_epochs: 20
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	6.25	100	2.5453	1.0
No log	12.5	200	1.9038	1.0
No log	18.75	300	1.3847	0.9798

Framework versions

Transformers 4.41.2
Pytorch 2.3.1+cu121
Datasets 2.19.2
Tokenizers 0.19.1