nkkbr/whisper-large-v3-zatoichi-ja-JDG_ver_20260212_lr_1.0e-5

Whisper Large v3 - Japanese Zatoichi ASR

This model is a fine-tuned version of openai/whisper-large-v3 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4799
• Wer: 66.7281
• Cer: 20.3677

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: 32
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
1.0467	0.0164	1	1.2545	98.1567	40.8419
1.1366	0.0328	2	0.9999	98.1567	39.5897
0.9644	0.0492	3	0.9732	95.2995	37.8447
0.8464	0.0656	4	0.9347	92.3502	36.4859
0.8562	0.0820	5	0.9078	88.0184	34.2480
0.8261	0.0984	6	0.8493	87.7419	32.5430
0.8442	0.1148	7	0.8080	85.9908	30.6514
0.855	0.1311	8	0.7666	90.6912	36.9122
0.6344	0.1475	9	0.7408	93.8249	38.5906
0.5937	0.1639	10	0.7163	89.0323	33.7818
0.7446	0.1803	11	0.6943	97.4194	38.9104
0.7102	0.1967	12	0.6736	102.5806	38.9104
0.7253	0.2131	13	0.6517	102.1198	35.2604
0.6416	0.2295	14	0.6361	106.6359	37.0454
0.579	0.2459	15	0.6232	99.6313	33.7019
0.564	0.2623	16	0.6161	80.9217	26.8416
0.4876	0.2787	17	0.6144	76.5899	25.7093
0.4848	0.2951	18	0.6073	80.5530	26.4420
0.5647	0.3115	19	0.6012	84.7926	27.5743
0.4575	0.3279	20	0.5974	76.5899	25.3497
0.5235	0.3443	21	0.5887	83.0415	27.9073
0.534	0.3607	22	0.5797	83.5023	28.0671
0.5456	0.3770	23	0.5837	77.9724	29.2261
0.539	0.3934	24	0.5811	88.2028	32.9825
0.55	0.4098	25	0.5747	81.1982	28.9330
0.4608	0.4262	26	0.5680	72.0737	24.3772
0.4046	0.4426	27	0.5640	72.8111	23.3382
0.4844	0.4590	28	0.5579	71.4286	22.9652
0.4507	0.4754	29	0.5597	74.1014	23.5913
0.4356	0.4918	30	0.5550	75.3917	24.1108
0.5625	0.5082	31	0.5482	73.0876	24.4971
0.5082	0.5246	32	0.5461	73.0876	23.8044
0.4272	0.5410	33	0.5439	78.8018	25.6028
0.573	0.5574	34	0.5408	78.5253	25.3763
0.4715	0.5738	35	0.5374	77.9724	25.8559
0.4589	0.5902	36	0.5327	77.3272	24.9634
0.4432	0.6066	37	0.5282	77.7880	25.0300
0.5121	0.6230	38	0.5254	78.3410	25.0833
0.4767	0.6393	39	0.5233	84.6083	29.7456
0.5307	0.6557	40	0.5215	73.7327	23.8577
0.439	0.6721	41	0.5192	71.3364	22.5256
0.4844	0.6885	42	0.5155	71.3364	22.6722
0.4597	0.7049	43	0.5120	70.6912	22.1393
0.4511	0.7213	44	0.5084	70.1382	21.4999
0.397	0.7377	45	0.5051	70.0461	21.8862
0.4419	0.7541	46	0.5028	70.7834	21.9262
0.495	0.7705	47	0.5011	70.5991	22.0061
0.4798	0.7869	48	0.5000	70.5069	22.0727
0.4115	0.8033	49	0.4990	70.1382	21.8596
0.5138	0.8197	50	0.4993	70.3226	22.8453
0.475	0.8361	51	0.4987	70.3226	22.0461
0.3971	0.8525	52	0.4967	71.0599	22.2725
0.5713	0.8689	53	0.4945	70.0461	22.0328
0.3532	0.8852	54	0.4923	70.6912	22.4457
0.4685	0.9016	55	0.4909	70.5991	22.0994
0.4398	0.9180	56	0.4908	70.7834	22.1926
0.3495	0.9344	57	0.4921	69.3088	22.0194
0.4142	0.9508	58	0.4911	72.9032	23.5380
0.428	0.9672	59	0.4888	73.1797	24.6437
0.4455	0.9836	60	0.4880	74.0092	23.7911
0.3517	1.0	61	0.4894	74.2857	23.7512
0.3689	1.0164	62	0.4902	70.2304	22.4990
0.2814	1.0328	63	0.4901	70.1382	23.2317
0.2134	1.0492	64	0.4904	69.7696	22.0461
0.2925	1.0656	65	0.4917	68.3871	21.4200
0.248	1.0820	66	0.4934	67.3733	21.0870
0.2614	1.0984	67	0.4967	66.9124	21.0470
0.315	1.1148	68	0.4954	67.3733	21.2468
0.3476	1.1311	69	0.4940	68.0184	21.3534
0.3538	1.1475	70	0.4923	67.6498	21.1403
0.2905	1.1639	71	0.4912	67.6498	21.1403
0.2355	1.1803	72	0.4917	67.8341	21.1936
0.2917	1.1967	73	0.4927	68.2949	21.2335
0.2742	1.2131	74	0.4937	68.2028	21.0071
0.2491	1.2295	75	0.4943	68.8479	21.0737
0.3151	1.2459	76	0.4939	68.6636	21.0337
0.2909	1.2623	77	0.4925	69.0323	21.1136
0.3073	1.2787	78	0.4909	68.7558	21.0737
0.2717	1.2951	79	0.4902	68.7558	20.9005
0.2548	1.3115	80	0.4897	68.3871	20.5941
0.268	1.3279	81	0.4890	68.0184	20.3810
0.2937	1.3443	82	0.4880	68.5714	20.6341
0.2956	1.3607	83	0.4870	68.4793	20.5808
0.2767	1.3770	84	0.4858	68.0184	20.3810
0.2506	1.3934	85	0.4850	68.5714	20.3810
0.2606	1.4098	86	0.4850	73.8249	22.1793
0.2646	1.4262	87	0.4850	74.0092	22.0727
0.2633	1.4426	88	0.4842	72.9032	22.0461
0.2846	1.4590	89	0.4849	72.9032	22.0194
0.2616	1.4754	90	0.4847	72.9954	22.0061
0.2929	1.4918	91	0.4843	73.0876	21.9662
0.258	1.5082	92	0.4844	67.9263	20.1945
0.1866	1.5246	93	0.4845	67.9263	20.2611
0.2814	1.5410	94	0.4840	67.7419	20.1545
0.2679	1.5574	95	0.4835	67.4654	20.2078
0.3164	1.5738	96	0.4830	67.1889	20.1678
0.2985	1.5902	97	0.4823	67.3733	20.2478
0.2711	1.6066	98	0.4815	67.1889	20.2211
0.2187	1.6230	99	0.4806	67.6498	20.5275
0.2406	1.6393	100	0.4799	67.4654	20.3543
0.2208	1.6557	101	0.4802	67.7419	20.3543
0.2527	1.6721	102	0.4798	67.4654	20.3011
0.2723	1.6885	103	0.4796	67.5576	20.3810
0.2886	1.7049	104	0.4793	67.4654	20.3011
0.3275	1.7213	105	0.4797	67.4654	20.3011
0.2476	1.7377	106	0.4794	67.4654	20.4209
0.2528	1.7541	107	0.4795	67.4654	20.3943
0.2302	1.7705	108	0.4795	66.9124	20.3810
0.2401	1.7869	109	0.4795	66.6359	20.1812
0.2608	1.8033	110	0.4797	66.6359	20.2344
0.3005	1.8197	111	0.4798	66.9124	20.3011
0.2266	1.8361	112	0.4798	66.7281	20.3011
0.228	1.8525	113	0.4800	66.4516	20.2744
0.2323	1.8689	114	0.4801	66.9124	20.3277
0.2959	1.8852	115	0.4797	66.4516	20.2344
0.2626	1.9016	116	0.4798	66.7281	20.3943
0.283	1.9180	117	0.4796	66.7281	20.4076
0.2996	1.9344	118	0.4796	66.9124	20.4076
0.2564	1.9508	119	0.4794	66.8203	20.3810
0.2487	1.9672	120	0.4798	66.9124	20.3543
0.2609	1.9836	121	0.4799	66.8203	20.3810
0.2407	2.0	122	0.4799	66.7281	20.3677

Framework versions

• Transformers 4.57.3
• Pytorch 2.9.1+cu128
• Datasets 4.4.1
• Tokenizers 0.22.1