Finetuned Whisper large-v3-turbo0 for darija speech translation

This model is a fine-tuned version of openai/whisper-large-v3-turbo on the Darija-C dataset. It achieves the following results on the evaluation set:

Model description

More information needed

More information needed

More information needed

The following hyperparameters were used during training:

learning_rate: 1e-05
train_batch_size: 16
eval_batch_size: 8
seed: 42
optimizer: Use 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: 100
training_steps: 1000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Bleu
2.4675	2.2727	50	1.2036	0.2616
0.8385	4.5455	100	0.4859	0.5348
0.3255	6.8182	150	0.1448	0.7604
0.1266	9.0909	200	0.0588	0.8582
0.0654	11.3636	250	0.0501	0.8531
0.0427	13.6364	300	0.0336	0.8875
0.0315	15.9091	350	0.0209	0.9159
0.0188	18.1818	400	0.0214	0.8977
0.0193	20.4545	450	0.0103	0.9233
0.012	22.7273	500	0.0053	0.9329
0.0084	25.0	550	0.0089	0.9291
0.0072	27.2727	600	0.0028	0.9332
0.0031	29.5455	650	0.0042	0.9326
0.0032	31.8182	700	0.0045	0.9318
0.0026	34.0909	750	0.0005	0.9362
0.0008	36.3636	800	0.0004	0.9364
0.0006	38.6364	850	0.0003	0.9369
0.0003	40.9091	900	0.0003	0.9369
0.0003	43.1818	950	0.0003	0.9369
0.0003	45.4545	1000	0.0003	0.9369