deepdml/whisper-tiny-ig-mix-norm

Whisper Tiny ig

This model is a fine-tuned version of openai/whisper-tiny on the google/fleurs dataset. It achieves the following results on the evaluation set:

• Loss: 1.2453
• Wer: 60.3907
• Cer: 25.7029

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-05
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.04
• training_steps: 5000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.2001	1.0406	1000	0.9724	58.8455	24.7463
0.0942	2.0812	2000	1.0967	60.7065	24.6946
0.0633	3.1218	3000	1.1767	59.6472	24.2311
0.0417	5.003	4000	1.2294	60.0262	25.0141
0.0392	6.0436	5000	1.2453	60.3907	25.7029

Framework versions

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

Citation

Please cite the model using the following BibTeX entry:

@misc{deepdml/whisper-tiny-ig-mix-norm,
      title={Fine-tuned Whisper tiny ASR model for speech recognition in Lingala},
      author={Jimenez, David},
      howpublished={\url{https://huggingface.co/deepdml/whisper-tiny-ig-mix-norm}},
      year={2025}
    }