This model is a fine-tuned version of openai/whisper-large-v3-turbo on the ntnu-smil/lttc-rebalanced-1-split dataset. It achieves the following results on the evaluation set:

Loss: 3.7199
Wer: 36.5280
Cer: 25.0

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: 0.0005
train_batch_size: 4
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
num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.0489	1.0	18	3.2509	37.2514	23.4504
0.1246	2.0	36	3.6744	35.6239	23.5709
0.0011	3.0	54	3.6182	36.7089	22.9855
0.0075	4.0	72	3.7182	37.1609	22.6240
0.0002	5.0	90	3.7643	37.7939	23.6398
0.0028	6.0	108	3.6117	36.7089	23.8809
0.0003	7.0	126	3.5535	36.8897	24.6556
0.0001	8.0	144	3.6586	37.7939	25.1033
0.0003	9.0	162	3.6168	36.8897	24.7934
0.0001	10.0	180	3.6500	37.1609	25.1033
0.0002	11.0	198	3.6934	37.4322	25.3960
0.0001	12.0	216	3.6901	36.9801	25.2410
0.0001	13.0	234	3.6980	36.7993	25.2238
0.0001	14.0	252	3.6990	36.9801	25.1377
0.0002	15.0	270	3.7110	36.9801	25.2755
0.0001	16.0	288	3.7139	36.7993	25.1894
0.0001	17.0	306	3.7175	36.7089	25.1722
0.0001	18.0	324	3.7202	36.9801	25.3444
0.0001	19.0	342	3.7210	36.8897	24.9828
0.0002	20.0	360	3.7199	36.5280	25.0

Framework versions

PEFT 0.13.2
Transformers 4.46.3
Pytorch 2.2.0+cu121
Datasets 3.1.0
Tokenizers 0.20.3

ntnu-smil/whisper-large-v3-turbo-score-5-rebalanced-2-merged

whisper-large-v3-turbo-score-5-rebalanced-2

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

ntnu-smil/
whisper-large-v3-turbo-score-5-rebalanced-2-merged