cantillation/Teamim-large-v3-turbo_WeightDecay-0.005_Augmented_WithSRT_date-15-04-2025

he-cantillation

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

• Loss: 0.4313
• Wer: 37.8893
• Avg Precision Exact: 0.4754
• Avg Recall Exact: 0.4866
• Avg F1 Exact: 0.4799
• Avg Precision Letter Shift: 0.4984
• Avg Recall Letter Shift: 0.5113
• Avg F1 Letter Shift: 0.5030
• Avg Precision Word Level: 0.5105
• Avg Recall Word Level: 0.5244
• Avg F1 Word Level: 0.5151
• Avg Precision Word Shift: 0.6800
• Avg Recall Word Shift: 0.7089
• Avg F1 Word Shift: 0.6906
• Precision Median Exact: 0.375
• Recall Median Exact: 0.4068
• F1 Median Exact: 0.3905
• 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	6.2393	110.8685	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
0.1465	0.2962	2500	0.7718	61.2193	0.2875	0.3030	0.2936	0.3219	0.3423	0.3297	0.3405	0.3623	0.3483	0.5405	0.5932	0.5599	0.1368	0.1579	0.1455	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
0.0667	0.5925	5000	0.6041	52.8951	0.3657	0.3756	0.3693	0.3989	0.4116	0.4035	0.4161	0.4290	0.4206	0.6173	0.6455	0.6280	0.2160	0.2333	0.2222	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
0.0576	0.8887	7500	0.5474	49.1429	0.3890	0.3981	0.3923	0.4180	0.4302	0.4223	0.4327	0.4457	0.4370	0.6206	0.6476	0.6303	0.2261	0.2428	0.2353	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
0.0552	1.1850	10000	0.5546	49.3049	0.3822	0.3988	0.3883	0.4092	0.4295	0.4165	0.4261	0.4496	0.4338	0.6053	0.6561	0.6226	0.2222	0.2576	0.2363	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
0.0497	1.4812	12500	0.5479	49.9147	0.3636	0.3712	0.3666	0.3935	0.4041	0.3975	0.4113	0.4225	0.4152	0.6057	0.6373	0.6173	0.2	0.2094	0.2013	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
0.0295	1.7775	15000	0.5342	48.7155	0.4055	0.4138	0.4083	0.4338	0.4439	0.4371	0.4500	0.4618	0.4535	0.6309	0.6577	0.6396	0.2470	0.2571	0.25	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
0.0377	2.0737	17500	0.5136	46.1304	0.3919	0.4037	0.3967	0.4175	0.4322	0.4232	0.4320	0.4483	0.4378	0.6157	0.6509	0.6288	0.2099	0.2308	0.2175	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
0.0202	2.3699	20000	0.5309	44.3608	0.4012	0.4122	0.4054	0.4276	0.4412	0.4325	0.4436	0.4596	0.4488	0.6299	0.6650	0.6427	0.2329	0.2576	0.2449	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
0.029	2.6662	22500	0.5077	46.8131	0.4103	0.4176	0.4129	0.4351	0.4440	0.4382	0.4502	0.4599	0.4530	0.6307	0.6557	0.6394	0.2275	0.25	0.2338	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
0.0124	2.9624	25000	0.4679	43.1952	0.4293	0.4428	0.4344	0.4533	0.4699	0.4596	0.4672	0.4853	0.4735	0.6417	0.6792	0.6556	0.3	0.3333	0.3114	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
0.0114	3.2587	27500	0.4747	47.7293	0.4276	0.4394	0.4321	0.4527	0.4666	0.4579	0.4666	0.4807	0.4714	0.6385	0.6666	0.6487	0.2812	0.3114	0.2963	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
0.0265	3.5549	30000	0.4438	42.8144	0.4250	0.4368	0.4294	0.4479	0.4631	0.4535	0.4625	0.4799	0.4683	0.6470	0.6832	0.6599	0.2778	0.3012	0.2857	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
0.0238	3.8512	32500	0.4454	41.6590	0.4508	0.4614	0.4549	0.4731	0.4859	0.4781	0.4868	0.5000	0.4919	0.6597	0.6896	0.6712	0.3191	0.3470	0.3276	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
0.0178	4.1474	35000	0.4334	40.8610	0.4631	0.4709	0.4661	0.4880	0.4972	0.4912	0.5021	0.5124	0.5053	0.6759	0.6998	0.6842	0.3401	0.3581	0.3478	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
0.0094	4.4437	37500	0.4594	42.0208	0.4477	0.4581	0.4516	0.4716	0.4853	0.4764	0.4855	0.5001	0.4903	0.6606	0.6936	0.6719	0.3125	0.3333	0.3182	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
0.008	4.7399	40000	0.4448	40.1024	0.4471	0.4582	0.4517	0.4702	0.4837	0.4754	0.4829	0.4978	0.4882	0.6558	0.6893	0.6685	0.3061	0.3333	0.3142	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
0.0167	5.0361	42500	0.4367	42.5547	0.4564	0.4656	0.4600	0.4798	0.4902	0.4836	0.4923	0.5035	0.4964	0.6707	0.6940	0.6789	0.3333	0.3539	0.3422	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
0.0028	5.3324	45000	0.4516	39.7654	0.4588	0.4687	0.4626	0.4807	0.4933	0.4855	0.4938	0.5078	0.4989	0.6760	0.7062	0.6872	0.3333	0.3600	0.3453	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
0.0055	5.6286	47500	0.4354	38.9630	0.4715	0.4809	0.4752	0.4942	0.5057	0.4982	0.5061	0.5197	0.5104	0.6761	0.7034	0.6856	0.3556	0.3801	0.3636	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
0.0196	5.9249	50000	0.4274	38.3824	0.4726	0.4855	0.4774	0.4945	0.5098	0.4999	0.5068	0.5234	0.5122	0.6692	0.7027	0.6811	0.375	0.4054	0.3823	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
0.0038	6.2211	52500	0.4482	38.9543	0.4652	0.4763	0.4699	0.4880	0.5013	0.4935	0.5008	0.5155	0.5065	0.6727	0.7037	0.6846	0.3552	0.3846	0.3691	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
0.0108	6.5174	55000	0.4453	39.4109	0.4742	0.4842	0.4783	0.4961	0.5079	0.5006	0.5083	0.5215	0.5128	0.6771	0.7048	0.6869	0.3624	0.3939	0.3787	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
0.0078	6.8136	57500	0.4373	38.4320	0.4784	0.4893	0.4828	0.5000	0.5133	0.5048	0.5132	0.5276	0.5180	0.6806	0.7116	0.6918	0.3797	0.4138	0.4000	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
0.0124	7.1098	60000	0.4313	37.8893	0.4754	0.4866	0.4799	0.4984	0.5113	0.5030	0.5105	0.5244	0.5151	0.6800	0.7089	0.6906	0.375	0.4068	0.3905	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 2.6.0+cu126
• Datasets 2.12.0
• Tokenizers 0.20.1