ndizi_whisper_large_v3_turbo_merged_may_26

This model is a fine-tuned version of openai/whisper-large-v3-turbo on the None 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: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 16
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
num_epochs: 25.0
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer	Cer	Score
1.9152	1.0	214	0.5378	0.4097	0.1780	70.6153
1.3513	2.0	428	0.5088	0.3678	0.1467	74.2775
0.9402	3.0	642	0.5202	0.3673	0.1325	75.0088
0.7373	4.0	856	0.5334	0.3613	0.1533	74.2683
0.4887	5.0	1070	0.5470	0.3627	0.1511	74.3126
0.3609	6.0	1284	0.5781	0.3473	0.1318	76.0417
0.2546	7.0	1498	0.6239	0.3455	0.1328	76.0830
0.1636	8.0	1712	0.6403	0.3385	0.1372	76.2142
0.1325	9.0	1926	0.6784	0.3500	0.1382	75.5893
0.0921	10.0	2140	0.6887	0.3340	0.1261	76.9922
0.0681	11.0	2354	0.7226	0.3302	0.1241	77.2852
0.0555	12.0	2568	0.7358	0.3340	0.1253	77.0320
0.0514	13.0	2782	0.7403	0.3426	0.1344	76.1504
0.0338	14.0	2996	0.7632	0.3277	0.1247	77.3776
0.0265	15.0	3210	0.7862	0.3374	0.1296	76.6513
0.0200	16.0	3424	0.7823	0.3539	0.1428	75.1636
0.0150	17.0	3638	0.8292	0.3301	0.1254	77.2277
0.0129	18.0	3852	0.8225	0.3292	0.1261	77.2337
0.0040	19.0	4066	0.8484	0.3237	0.1230	77.6658
0.0042	20.0	4280	0.8747	0.3260	0.1222	77.5902
0.0032	21.0	4494	0.8716	0.3323	0.1311	76.8293
0.0022	22.0	4708	0.8932	0.3236	0.1229	77.6743
0.0015	23.0	4922	0.9002	0.3252	0.1236	77.5642
0.0010	24.0	5136	0.9021	0.3241	0.1237	77.6107
0.0008	25.0	5350	0.9059	0.3238	0.1238	77.6170