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: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.3462	0.0352	5	1.3332	97.6657
1.2744	0.0704	10	1.2213	122.7178
1.053	0.1056	15	1.0540	140.3501
0.9646	0.1408	20	0.9289	137.7241
0.9017	0.1761	25	0.8334	107.8366
0.7724	0.2113	30	0.7862	90.3710
0.7219	0.2465	35	0.7459	77.8658
0.6191	0.2817	40	0.7193	74.4894
0.6083	0.3169	45	0.6992	76.0734
0.6737	0.3521	50	0.6746	71.1963
0.6233	0.3873	55	0.6549	93.2889
0.6872	0.4225	60	0.6473	76.7403
0.5576	0.4577	65	0.6257	87.9950
0.5729	0.4930	70	0.6114	83.9516
0.5057	0.5282	75	0.6060	79.1580
0.6545	0.5634	80	0.5893	71.1963
0.4883	0.5986	85	0.5800	82.1175
0.5538	0.6338	90	0.5765	83.9933
0.5893	0.6690	95	0.5620	73.6557
0.5145	0.7042	100	0.5583	76.2818
0.4707	0.7394	105	0.5577	74.7395
0.5054	0.7746	110	0.5430	65.4439
0.5323	0.8099	115	0.5318	71.0713
0.5412	0.8451	120	0.5360	78.6578
0.4255	0.8803	125	0.5229	70.1959
0.4482	0.9155	130	0.5174	74.8645
0.4819	0.9507	135	0.5099	86.6194
0.4457	0.9859	140	0.5060	67.7366
0.3657	1.0211	145	0.5015	67.1530
0.1885	1.0563	150	0.5023	65.9858
0.3217	1.0915	155	0.4992	65.4023
0.2822	1.1268	160	0.4968	67.4031
0.3064	1.1620	165	0.4987	73.4473
0.4926	1.1972	170	0.4892	64.0684
0.2883	1.2324	175	0.4925	69.1121
0.3175	1.2676	180	0.4913	58.8579
0.3407	1.3028	185	0.4788	59.9000
0.3021	1.3380	190	0.4744	60.2334
0.255	1.3732	195	0.4659	58.7328
0.2961	1.4085	200	0.4665	62.4844
0.2944	1.4437	205	0.4699	63.8183
0.2986	1.4789	210	0.4604	61.1922
0.3836	1.5141	215	0.4595	67.7782
0.2638	1.5493	220	0.4558	62.9846
0.2811	1.5845	225	0.4520	69.3622
0.3104	1.6197	230	0.4465	58.8995
0.4079	1.6549	235	0.4483	59.8166
0.2565	1.6901	240	0.4397	58.0659
0.2719	1.7254	245	0.4357	62.1926
0.3146	1.7606	250	0.4345	61.8174
0.383	1.7958	255	0.4288	53.5223
0.3292	1.8310	260	0.4268	59.2330
0.2884	1.8662	265	0.4195	59.9000
0.184	1.9014	270	0.4200	63.1513
0.3253	1.9366	275	0.4194	66.7778
0.2564	1.9718	280	0.4157	60.4002
0.2414	2.0070	285	0.4124	58.5244
0.1988	2.0423	290	0.4164	56.6486
0.136	2.0775	295	0.4247	57.3155
0.1716	2.1127	300	0.4388	59.9000
0.286	2.1479	305	0.4377	58.4827
0.2436	2.1831	310	0.4288	58.9829
0.2296	2.2183	315	0.4227	60.4002
0.1302	2.2535	320	0.4234	51.0629
0.1275	2.2887	325	0.4233	53.1888
0.1387	2.3239	330	0.4198	50.7295
0.1193	2.3592	335	0.4229	51.1046
0.147	2.3944	340	0.4149	51.2297
0.1881	2.4296	345	0.4105	48.8537
0.145	2.4648	350	0.4017	49.4789
0.0996	2.5	355	0.4010	49.8958
0.1206	2.5352	360	0.4115	54.5227
0.1274	2.5704	365	0.4079	48.9371
0.0974	2.6056	370	0.4092	49.5206
0.1456	2.6408	375	0.4113	46.9362
0.1485	2.6761	380	0.4052	54.9396
0.1349	2.7113	385	0.4012	51.4798
0.1512	2.7465	390	0.4014	51.3964
0.103	2.7817	395	0.3991	47.1863
0.1698	2.8169	400	0.3964	46.9779
0.152	2.8521	405	0.3937	47.5615
0.1134	2.8873	410	0.3902	46.2693
0.1742	2.9225	415	0.3897	46.1859
0.1424	2.9577	420	0.3926	48.4368
0.0846	2.9930	425	0.3918	50.1459
0.072	3.0282	430	0.3945	45.6857
0.0646	3.0634	435	0.4046	46.3943
0.101	3.0986	440	0.4195	47.1446
0.0511	3.1338	445	0.4205	45.0188
0.1047	3.1690	450	0.4200	45.0604
0.0744	3.2042	455	0.4162	44.5602
0.0354	3.2394	460	0.4146	44.4769
0.0979	3.2746	465	0.4137	44.3518
0.0738	3.3099	470	0.4124	43.9767
0.0593	3.3451	475	0.4165	44.6436
0.073	3.3803	480	0.4155	44.5602
0.0679	3.4155	485	0.4135	43.9350
0.0664	3.4507	490	0.4134	43.3514
0.0456	3.4859	495	0.4146	44.7270
0.0542	3.5211	500	0.4137	43.8099
0.0803	3.5563	505	0.4145	44.1017
0.0597	3.5915	510	0.4121	47.2280
0.0682	3.6268	515	0.4096	46.3110
0.0752	3.6620	520	0.4082	46.5611
0.1061	3.6972	525	0.4065	46.7695
0.0461	3.7324	530	0.4065	43.5181
0.0421	3.7676	535	0.4061	43.2680
0.0905	3.8028	540	0.4065	43.5181
0.0481	3.8380	545	0.4068	43.2263
0.0612	3.8732	550	0.4072	43.1430
0.0979	3.9085	555	0.4071	43.3931
0.0552	3.9437	560	0.4066	43.0596
0.0746	3.9789	565	0.4067	42.8095

Framework versions

Transformers 4.57.3
Pytorch 2.9.1+cu128
Datasets 4.4.1
Tokenizers 0.22.1

nkkbr/whisper-large-v3-zatoichi-ja-EX-7

Whisper Large v3 - Japanese Zatoichi ASR

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

nkkbr/
whisper-large-v3-zatoichi-ja-EX-7