whisper-tiny-swc-drc-kat-2

This model is a fine-tuned version of openai/whisper-medium on the Ussen/swc-drc-katanga dataset. It achieves the following results on the evaluation set:

• Loss: 0.8770
• Wer Ortho: 39.7101
• Wer: 0.3914

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 500
• training_steps: 4000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
0.3735	2.96	1000	0.6459	48.5374	0.4783
0.081	5.93	2000	0.7478	38.5969	0.3797
0.0279	8.89	3000	0.8653	38.7523	0.3813
0.0189	11.85	4000	0.8770	39.7101	0.3914

Framework versions

• Transformers 4.33.0.dev0
• Pytorch 1.13.1+cu117
• Datasets 2.13.1
• Tokenizers 0.12.1