Whisper Tunisien

This model is a fine-tuned version of openai/whisper-small on the Tunisian_dataset_STT-TTS1 dataset. It achieves the following results on the evaluation set:

Loss: 2.8010
Wer: 102.9831

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: 1e-07
train_batch_size: 8
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 5000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.1413	10.3093	1000	2.9765	135.4086
1.0017	20.6186	2000	2.8593	107.4578
0.9227	30.9278	3000	2.8201	100.3891
0.8583	41.2371	4000	2.8054	102.9831
0.8442	51.5464	5000	2.8010	102.9831

Framework versions

Transformers 4.42.0.dev0
Pytorch 2.3.0+cu121
Datasets 2.19.1
Tokenizers 0.19.1

Downloads last month: 5

Safetensors

Model size

0.2B params

Tensor type

F32

Model tree for Arbi-Houssem/TunLangModel1.7

Base model

openai/whisper-small

Finetuned

(3799)

this model

Dataset used to train Arbi-Houssem/TunLangModel1.7

Evaluation results

Wer on Tunisian_dataset_STT-TTS1
self-reported

102.983