sheickydollar/SaChi-ASR

SaCHi_ASR

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

• Loss: 0.2234
• Wer: 21.3111

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 250
• training_steps: 2000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4253	0.4657	400	0.4046	34.3409
0.2789	0.9313	800	0.2871	24.3238
0.2239	1.3970	1200	0.2498	19.3530
0.2122	1.8626	1600	0.2270	23.5416
0.1209	2.3283	2000	0.2234	21.3111

Framework versions

• Transformers 4.50.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1