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
1.6078	0.0391	5	1.3113	94.5968
1.2834	0.0781	10	1.1757	116.8548
1.2326	0.1172	15	1.0430	124.3548
0.9735	0.1562	20	0.9291	101.6935
0.9705	0.1953	25	0.8421	96.2903
0.9142	0.2344	30	0.8056	93.4677
0.9752	0.2734	35	0.7907	93.1452
0.9126	0.3125	40	0.7571	85.0806
0.8125	0.3516	45	0.7777	90.3226
0.8412	0.3906	50	0.7349	80.0
0.7375	0.4297	55	0.7412	100.6452
0.7459	0.4688	60	0.7253	85.5645
0.8004	0.5078	65	0.7177	77.2581
0.7954	0.5469	70	0.7405	108.4677
0.5945	0.5859	75	0.7050	88.3871
0.7378	0.625	80	0.6988	78.2258
0.7327	0.6641	85	0.7242	90.1613
0.6521	0.7031	90	0.6941	80.2419
0.6377	0.7422	95	0.6843	79.8387
0.5972	0.7812	100	0.6864	84.3548
0.7501	0.8203	105	0.6876	78.8710
0.6684	0.8594	110	0.6916	89.0323
0.7048	0.8984	115	0.6723	80.3226
0.6409	0.9375	120	0.6656	79.2742
0.5408	0.9766	125	0.6757	86.2903
0.4808	1.0156	130	0.6747	82.7419
0.4608	1.0547	135	0.6666	80.2419
0.4305	1.0938	140	0.6838	80.0
0.4927	1.1328	145	0.6858	81.9355
0.3838	1.1719	150	0.6866	80.0
0.4763	1.2109	155	0.6881	90.5645
0.3953	1.25	160	0.6859	87.7419
0.5002	1.2891	165	0.6896	103.7097
0.4152	1.3281	170	0.6732	86.8548
0.3067	1.3672	175	0.6777	87.0968
0.4965	1.4062	180	0.6966	101.9355
0.4117	1.4453	185	0.6639	84.0323
0.3928	1.4844	190	0.6630	85.8871
0.555	1.5234	195	0.6757	93.5484
0.4058	1.5625	200	0.6586	78.3065
0.3633	1.6016	205	0.6630	89.3548
0.4365	1.6406	210	0.6846	88.5484
0.4667	1.6797	215	0.6612	80.8871
0.4338	1.7188	220	0.6624	76.2903
0.3751	1.7578	225	0.6697	81.6935
0.4858	1.7969	230	0.6620	79.5968
0.4232	1.8359	235	0.6513	79.2742
0.4217	1.875	240	0.6525	78.8710
0.5218	1.9141	245	0.6625	80.3226
0.5	1.9531	250	0.6575	81.2097
0.4541	1.9922	255	0.6556	81.5323
0.3422	2.0312	260	0.6578	81.6129
0.2671	2.0703	265	0.6659	79.2742
0.2297	2.1094	270	0.6967	83.0645
0.2592	2.1484	275	0.7118	78.9516
0.1961	2.1875	280	0.7059	75.5645
0.2508	2.2266	285	0.7028	80.2419
0.2125	2.2656	290	0.7144	83.5484
0.2409	2.3047	295	0.7029	79.7581
0.1756	2.3438	300	0.6951	77.9032
0.2416	2.3828	305	0.7053	83.5484
0.2144	2.4219	310	0.6899	87.1774
0.2296	2.4609	315	0.6904	79.5161
0.1785	2.5	320	0.7114	85.0
0.235	2.5391	325	0.7120	81.8548
0.2378	2.5781	330	0.7064	85.2419
0.3536	2.6172	335	0.7096	89.1935
0.2343	2.6562	340	0.7153	85.6452
0.2169	2.6953	345	0.6933	79.1935
0.2089	2.7344	350	0.6917	88.8710
0.2791	2.7734	355	0.7063	86.2903
0.2185	2.8125	360	0.7061	83.0645
0.2382	2.8516	365	0.6987	81.0484
0.2602	2.8906	370	0.6965	79.5161
0.2145	2.9297	375	0.7010	81.1290
0.2836	2.9688	380	0.6993	82.8226
0.1479	3.0078	385	0.6922	80.7258
0.1075	3.0469	390	0.6995	79.4355
0.1336	3.0859	395	0.7174	89.3548
0.131	3.125	400	0.7327	81.0484
0.2023	3.1641	405	0.7435	79.1935
0.1218	3.2031	410	0.7557	85.3226
0.0988	3.2422	415	0.7550	82.0161
0.1353	3.2812	420	0.7461	81.1290
0.182	3.3203	425	0.7526	82.9839
0.1905	3.3594	430	0.7573	83.5484
0.1253	3.3984	435	0.7540	82.6613
0.1215	3.4375	440	0.7461	77.3387
0.1516	3.4766	445	0.7497	80.2419
0.1082	3.5156	450	0.7618	82.9839
0.15	3.5547	455	0.7580	82.0161
0.0802	3.5938	460	0.7514	80.8065
0.1185	3.6328	465	0.7498	80.0806
0.1649	3.6719	470	0.7528	80.8065
0.1537	3.7109	475	0.7570	82.3387
0.0595	3.75	480	0.7566	82.6613
0.1907	3.7891	485	0.7560	81.2097
0.0934	3.8281	490	0.7560	80.4839
0.1398	3.8672	495	0.7568	80.1613
0.1348	3.9062	500	0.7566	80.2419
0.1309	3.9453	505	0.7569	80.5645
0.0954	3.9844	510	0.7573	80.5645

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-1-TRAIN_2_TO_10_EVAL_1

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-1-TRAIN_2_TO_10_EVAL_1