he-cantillation

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

Loss: 0.2584
Wer: 17.4613
Avg Precision Exact: 0.8722
Avg Recall Exact: 0.8838
Avg F1 Exact: 0.8772
Avg Precision Letter Shift: 0.8981
Avg Recall Letter Shift: 0.9102
Avg F1 Letter Shift: 0.9033
Avg Precision Word Level: 0.9008
Avg Recall Word Level: 0.9121
Avg F1 Word Level: 0.9057
Avg Precision Word Shift: 0.9601
Avg Recall Word Shift: 0.9730
Avg F1 Word Shift: 0.9657
Precision Median Exact: 0.9286
Recall Median Exact: 0.9333
F1 Median Exact: 0.9565
Precision Max Exact: 1.0
Recall Max Exact: 1.0
F1 Max Exact: 1.0
Precision Min Exact: 0.0
Recall Min Exact: 0.0
F1 Min Exact: 0.0
Precision Min Letter Shift: 0.0
Recall Min Letter Shift: 0.0
F1 Min Letter Shift: 0.0
Precision Min Word Level: 0.0
Recall Min Word Level: 0.0
F1 Min Word Level: 0.0
Precision Min Word Shift: 0.6364
Recall Min Word Shift: 0.6364
F1 Min Word Shift: 0.6364

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: 8
eval_batch_size: 32
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 200000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Avg Precision Exact	Avg Recall Exact	Avg F1 Exact	Avg Precision Letter Shift	Avg Recall Letter Shift	Avg F1 Letter Shift	Avg Precision Word Level	Avg Recall Word Level	Avg F1 Word Level	Avg Precision Word Shift	Avg Recall Word Shift	Avg F1 Word Shift	Precision Median Exact	Recall Median Exact	F1 Median Exact	Precision Max Exact	Recall Max Exact	F1 Max Exact	Precision Min Word Shift	Recall Min Word Shift	F1 Min Word Shift
No log	0.0001	1	10.1840	103.2210	0.0002	0.0000	0.0000	0.0022	0.0069	0.0014	0.0019	0.0169	0.0033	0.0094	0.0316	0.0107	0.0	0.0	0.0	0.0714	0.0069	0.0127	0.0	0.0	0.0
0.0146	1.1457	10000	0.1688	23.0557	0.8202	0.8417	0.8300	0.8524	0.8746	0.8625	0.8584	0.8783	0.8674	0.9348	0.9545	0.9436	0.9091	0.9167	0.9091	1.0	1.0	1.0	0.2308	0.375	0.2857
0.0105	2.2915	20000	0.1836	20.5764	0.8526	0.8616	0.8563	0.8800	0.8897	0.8840	0.8816	0.8917	0.8858	0.9510	0.9624	0.9559	0.9167	0.9231	0.9231	1.0	1.0	1.0	0.5556	0.625	0.5882
0.0051	3.4372	30000	0.1949	19.7288	0.8570	0.8663	0.8608	0.8812	0.8912	0.8854	0.8834	0.8928	0.8872	0.9540	0.9642	0.9583	0.9231	0.9286	0.9474	1.0	1.0	1.0	0.6429	0.6923	0.6667
0.0025	4.5830	40000	0.2128	19.5380	0.8631	0.8765	0.8690	0.8908	0.9051	0.8971	0.8951	0.9065	0.9000	0.9514	0.9654	0.9577	0.9231	0.9231	0.9231	1.0	1.0	1.0	0.6364	0.6923	0.6667
0.003	5.7287	50000	0.2182	20.0042	0.8632	0.8761	0.8689	0.8901	0.9039	0.8962	0.8935	0.9057	0.8988	0.9539	0.9671	0.9597	0.9231	0.9231	0.9286	1.0	1.0	1.0	0.6154	0.6667	0.64
0.0027	6.8744	60000	0.2184	19.0083	0.8532	0.8641	0.8579	0.8810	0.8923	0.8859	0.8840	0.8941	0.8884	0.9564	0.9677	0.9613	0.9231	0.9231	0.9231	1.0	1.0	1.0	0.6667	0.6923	0.6923
0.0016	8.0202	70000	0.2156	18.9871	0.8527	0.8636	0.8574	0.8808	0.8923	0.8858	0.8825	0.8938	0.8874	0.9544	0.9653	0.9591	0.9167	0.9231	0.9231	1.0	1.0	1.0	0.6667	0.6667	0.6667
0.001	9.1659	80000	0.2248	18.9659	0.8610	0.8698	0.8647	0.8903	0.9000	0.8944	0.8933	0.9020	0.8969	0.9569	0.9673	0.9614	0.9231	0.9231	0.9231	1.0	1.0	1.0	0.4286	0.5	0.4615
0.001	10.3116	90000	0.2344	18.6692	0.8763	0.8880	0.8814	0.9011	0.9133	0.9063	0.9039	0.9152	0.9087	0.9558	0.9694	0.9618	0.9231	0.9286	0.9524	1.0	1.0	1.0	0.5833	0.6923	0.6667
0.0008	11.4574	100000	0.2362	18.8811	0.8607	0.8709	0.8650	0.8878	0.8983	0.8922	0.8905	0.8994	0.8942	0.9534	0.9654	0.9586	0.9231	0.9231	0.9286	1.0	1.0	1.0	0.6429	0.6923	0.6923
0.0005	12.6031	110000	0.2331	18.3513	0.8603	0.8711	0.8648	0.8878	0.8992	0.8926	0.8897	0.9010	0.8945	0.9516	0.9631	0.9566	0.9231	0.9231	0.9286	1.0	1.0	1.0	0.1667	0.2	0.1818
0.0002	13.7489	120000	0.2378	18.6692	0.8691	0.8790	0.8734	0.8960	0.9066	0.9006	0.8999	0.9102	0.9043	0.9583	0.9688	0.9628	0.9231	0.9231	0.9286	1.0	1.0	1.0	0.5455	0.5455	0.5455
0.0005	14.8946	130000	0.2428	18.3090	0.8702	0.8790	0.8739	0.8960	0.9053	0.8999	0.8997	0.9072	0.9027	0.9563	0.9671	0.9609	0.9231	0.9231	0.9524	1.0	1.0	1.0	0.6923	0.6923	0.6923
0.0001	16.0403	140000	0.2476	18.0123	0.8701	0.8823	0.8754	0.8959	0.9085	0.9014	0.8993	0.9109	0.9043	0.9584	0.9715	0.9641	0.9231	0.9286	0.9524	1.0	1.0	1.0	0.6923	0.6923	0.6923
0.0	17.1861	150000	0.2520	18.4361	0.8639	0.8777	0.8700	0.8910	0.9054	0.8974	0.8941	0.9079	0.9002	0.9590	0.9723	0.9648	0.9231	0.9286	0.9524	1.0	1.0	1.0	0.6923	0.7778	0.7500
0.0	18.3318	160000	0.2503	18.1394	0.8682	0.8791	0.8729	0.8945	0.9061	0.8995	0.8980	0.9083	0.9024	0.9596	0.9715	0.9648	0.9286	0.9286	0.9524	1.0	1.0	1.0	0.6923	0.75	0.7200
0.0	19.4775	170000	0.2474	17.8004	0.8607	0.8714	0.8653	0.8866	0.8977	0.8913	0.8900	0.9006	0.8945	0.9601	0.9701	0.9643	0.9286	0.9286	0.9565	1.0	1.0	1.0	0.6364	0.6364	0.6364
0.0	20.6233	180000	0.2517	17.9275	0.8658	0.8766	0.8704	0.8928	0.9040	0.8976	0.8955	0.9057	0.8998	0.9604	0.9722	0.9655	0.9231	0.9286	0.9524	1.0	1.0	1.0	0.6364	0.6364	0.6364
0.0	21.7690	190000	0.2553	17.6520	0.8667	0.8780	0.8716	0.8924	0.9041	0.8974	0.8950	0.9060	0.8997	0.9597	0.9718	0.9649	0.9231	0.9286	0.9524	1.0	1.0	1.0	0.6364	0.6364	0.6364
0.0	22.9148	200000	0.2584	17.4613	0.8722	0.8838	0.8772	0.8981	0.9102	0.9033	0.9008	0.9121	0.9057	0.9601	0.9730	0.9657	0.9286	0.9333	0.9565	1.0	1.0	1.0	0.6364	0.6364	0.6364

Framework versions

Transformers 4.41.2
Pytorch 2.2.1
Datasets 2.20.0
Tokenizers 0.19.1

cantillation
/

Teamim-small_WeightDecay-0.05_Augmented_Old-Data_date-21-07-2024_14-34_WithNikud

he-cantillation

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

Model tree for cantillation/Teamim-small_WeightDecay-0.05_Augmented_Old-Data_date-21-07-2024_14-34_WithNikud

Evaluation results