cantillation/Teamim-IvritAI-large-v3-turbo-new_WeightDecay-0.005_Augmented_WithSRT_date-23-04-2025

he-cantillation

This model is a fine-tuned version of ivrit-ai/whisper-large-v3-turbo on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4107
• Wer: 36.2146
• Avg Precision Exact: 0.4891
• Avg Recall Exact: 0.4995
• Avg F1 Exact: 0.4932
• Avg Precision Letter Shift: 0.5095
• Avg Recall Letter Shift: 0.5218
• Avg F1 Letter Shift: 0.5143
• Avg Precision Word Level: 0.5211
• Avg Recall Word Level: 0.5357
• Avg F1 Word Level: 0.5263
• Avg Precision Word Shift: 0.6939
• Avg Recall Word Shift: 0.7234
• Avg F1 Word Shift: 0.7047
• Precision Median Exact: 0.4432
• Recall Median Exact: 0.4756
• F1 Median Exact: 0.4528
• 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: 0.0
• Recall Min Word Shift: 0.0
• F1 Min Word Shift: 0.0

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 2
• 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: 1000
• training_steps: 60000
• 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 Exact	Recall Min Exact	F1 Min Exact	Precision Min Letter Shift	Recall Min Letter Shift	F1 Min Letter Shift	Precision Min Word Level	Recall Min Word Level	F1 Min Word Level	Precision Min Word Shift	Recall Min Word Shift	F1 Min Word Shift
No log	0.0001	1	7.1601	109.4023	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
0.1361	0.2962	2500	0.6842	62.4024	0.2897	0.2982	0.2919	0.3222	0.3329	0.3249	0.3400	0.3543	0.3446	0.5235	0.5550	0.5339	0.1505	0.1562	0.1538	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0633	0.5925	5000	0.5766	51.9556	0.3589	0.3642	0.3604	0.3909	0.3982	0.3927	0.4080	0.4161	0.4098	0.6144	0.6383	0.6218	0.2	0.2185	0.2122	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0629	0.8887	7500	0.5497	53.1810	0.3695	0.3923	0.3759	0.3997	0.4249	0.4062	0.4143	0.4424	0.4213	0.5923	0.6461	0.6078	0.2222	0.2639	0.2381	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0375	1.1850	10000	0.5146	45.5717	0.4020	0.4168	0.4084	0.4303	0.4478	0.4374	0.4451	0.4634	0.4521	0.6430	0.6809	0.6579	0.2571	0.2857	0.2675	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0513	1.4812	12500	0.4789	47.3763	0.4082	0.4225	0.4140	0.4363	0.4528	0.4429	0.4515	0.4685	0.4581	0.6408	0.6728	0.6531	0.2698	0.2941	0.2798	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0407	1.7775	15000	0.4574	45.7832	0.4313	0.4414	0.4353	0.4612	0.4728	0.4654	0.4754	0.4873	0.4795	0.6548	0.6794	0.6639	0.3333	0.3438	0.3393	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0228	2.0737	17500	0.4692	45.7453	0.4131	0.4225	0.4167	0.4392	0.4496	0.4427	0.4549	0.4671	0.4590	0.6411	0.6686	0.6508	0.2587	0.2759	0.2659	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0203	2.3699	20000	0.4669	43.5424	0.4200	0.4337	0.4250	0.4432	0.4599	0.4493	0.4563	0.4741	0.4622	0.6356	0.6750	0.6493	0.2795	0.3077	0.2919	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0469	2.6662	22500	0.4451	42.7925	0.4403	0.4521	0.4447	0.4662	0.4808	0.4714	0.4814	0.4968	0.4867	0.6632	0.6933	0.6740	0.3118	0.3333	0.3144	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0224	2.9624	25000	0.4347	42.4672	0.4382	0.4533	0.4443	0.4603	0.4782	0.4674	0.4721	0.4924	0.4799	0.6405	0.6819	0.6567	0.3204	0.3536	0.3312	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0435	3.2587	27500	0.4208	40.8420	0.4435	0.4582	0.4493	0.4658	0.4829	0.4724	0.4791	0.4971	0.4860	0.6601	0.6978	0.6746	0.3333	0.3529	0.3352	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0379	3.5549	30000	0.4347	42.3782	0.4409	0.4491	0.4438	0.4652	0.4750	0.4684	0.4796	0.4898	0.4822	0.6691	0.6910	0.6760	0.3066	0.3295	0.3136	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0099	3.8512	32500	0.4303	43.9873	0.4492	0.4603	0.4534	0.4726	0.4869	0.4776	0.4849	0.5005	0.4902	0.6510	0.6820	0.6619	0.3333	0.3571	0.3388	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0185	4.1474	35000	0.4299	41.5102	0.4465	0.4562	0.4502	0.4709	0.4819	0.4749	0.4856	0.4980	0.4898	0.6627	0.6909	0.6726	0.3239	0.3483	0.3333	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0077	4.4437	37500	0.4209	41.2943	0.4518	0.4630	0.4559	0.4754	0.4885	0.4800	0.4885	0.5024	0.4932	0.6663	0.6955	0.6756	0.3429	0.3684	0.3552	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0126	4.7399	40000	0.4175	38.7048	0.4651	0.4746	0.4687	0.4893	0.5012	0.4936	0.5021	0.5152	0.5069	0.6841	0.7125	0.6943	0.3704	0.3974	0.3851	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0035	5.0361	42500	0.4141	37.2328	0.4746	0.4863	0.4793	0.4966	0.5105	0.5021	0.5085	0.5241	0.5143	0.6849	0.7159	0.6969	0.3974	0.4231	0.4114	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0085	5.3324	45000	0.4234	40.2366	0.4689	0.4787	0.4727	0.4922	0.5037	0.4962	0.5055	0.5182	0.5093	0.6792	0.7057	0.6880	0.375	0.4054	0.3885	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0141	5.6286	47500	0.4107	38.0483	0.4664	0.4765	0.4703	0.4871	0.4991	0.4916	0.4988	0.5122	0.5033	0.6747	0.7043	0.6856	0.3675	0.3921	0.3846	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0026	5.9249	50000	0.4120	37.9521	0.4745	0.4862	0.4788	0.4953	0.5094	0.5004	0.5071	0.5226	0.5127	0.6811	0.7128	0.6928	0.4	0.4167	0.4065	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0059	6.2211	52500	0.4140	36.1839	0.4890	0.4993	0.4931	0.5123	0.5244	0.5167	0.5241	0.5372	0.5288	0.6995	0.7273	0.7099	0.4444	0.4674	0.4528	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0069	6.5174	55000	0.4083	37.5086	0.4888	0.4982	0.4924	0.5093	0.5203	0.5133	0.5221	0.5340	0.5261	0.6960	0.7203	0.7042	0.4329	0.4545	0.4430	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0042	6.8136	57500	0.4116	36.6785	0.4906	0.5013	0.4948	0.5110	0.5233	0.5157	0.5225	0.5356	0.5274	0.6950	0.7236	0.7059	0.4336	0.4615	0.4444	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.004	7.1098	60000	0.4107	36.2146	0.4891	0.4995	0.4932	0.5095	0.5218	0.5143	0.5211	0.5357	0.5263	0.6939	0.7234	0.7047	0.4432	0.4756	0.4528	1.0	1.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0

Framework versions

• Transformers 4.49.0
• Pytorch 2.7.0+cu126
• Datasets 2.12.0
• Tokenizers 0.20.1