MIX2_ja-en_helsinki

This model is a fine-tuned version of Helsinki-NLP/opus-mt-ja-en on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.4929
• Otaku Benchmark VN BLEU: 20.21
• Otaku Benchmark LN BLEU: 13.29
• Otaku Benchmark MANGA BLEU: 19.07

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: 0.0003
• train_batch_size: 96
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.8467	0.01	2000	2.3237
2.6439	0.02	4000	2.2542
2.547	0.03	6000	2.1956
2.4852	0.04	8000	2.1088
2.4408	0.05	10000	2.0909
2.404	0.06	12000	2.1029
2.3634	0.07	14000	2.0636
2.3491	0.08	16000	2.0312
2.3203	0.09	18000	2.0187
2.3002	0.1	20000	1.9999
2.2791	0.11	22000	1.9823
2.2607	0.11	24000	1.9588
2.2475	0.12	26000	1.9728
2.2308	0.13	28000	1.9330
2.2237	0.14	30000	1.9657
2.208	0.15	32000	1.9560
2.2019	0.16	34000	1.9704
2.1864	0.17	36000	1.9513
2.1764	0.18	38000	1.9534
2.163	0.19	40000	1.9140
2.1534	0.2	42000	1.9241
2.146	0.21	44000	1.9162
2.1403	0.22	46000	1.9030
2.1309	0.23	48000	1.8741
2.1174	0.24	50000	1.8834
2.1157	0.25	52000	1.8666
2.1116	0.26	54000	1.8870
2.1062	0.27	56000	1.8837
2.0994	0.28	58000	1.8638
2.0924	0.29	60000	1.8766
2.0874	0.3	62000	1.8712
2.0805	0.31	64000	1.8792
2.0746	0.32	66000	1.8586
2.0684	0.32	68000	1.8819
2.0678	0.33	70000	1.8529
2.061	0.34	72000	1.8219
2.0532	0.35	74000	1.8383
2.0536	0.36	76000	1.8273
2.0432	0.37	78000	1.8304
2.0386	0.38	80000	1.8208
2.0361	0.39	82000	1.8103
2.0353	0.4	84000	1.8193
2.0266	0.41	86000	1.8369
2.0277	0.42	88000	1.8266
2.0221	0.43	90000	1.8372
2.0181	0.44	92000	1.8436
2.0182	0.45	94000	1.8505
2.0088	0.46	96000	1.8127
2.005	0.47	98000	1.8325
2.0003	0.48	100000	1.8407
2.0031	0.49	102000	1.8140
1.9954	0.5	104000	1.8177
1.9894	0.51	106000	1.8072
1.9901	0.52	108000	1.7971
1.9864	0.53	110000	1.8007
1.9848	0.53	112000	1.7961
1.9774	0.54	114000	1.7933
1.9802	0.55	116000	1.8031
1.9698	0.56	118000	1.8137
1.973	0.57	120000	1.7930
1.9696	0.58	122000	1.7838
1.9641	0.59	124000	1.7730
1.9609	0.6	126000	1.7800
1.9605	0.61	128000	1.7680
1.9516	0.62	130000	1.7895
1.9529	0.63	132000	1.7825
1.9503	0.64	134000	1.7792
1.9528	0.65	136000	1.8031
1.9439	0.66	138000	1.7652
1.9453	0.67	140000	1.7713
1.9404	0.68	142000	1.7585
1.9399	0.69	144000	1.7454
1.9325	0.7	146000	1.7605
1.9327	0.71	148000	1.7608
1.9301	0.72	150000	1.7743
1.928	0.73	152000	1.7532
1.9286	0.74	154000	1.7682
1.9194	0.74	156000	1.7582
1.9247	0.75	158000	1.7601
1.9183	0.76	160000	1.7600
1.9138	0.77	162000	1.7555
1.9148	0.78	164000	1.7447
1.913	0.79	166000	1.7512
1.9084	0.8	168000	1.7408
1.9109	0.81	170000	1.7463
1.905	0.82	172000	1.7543
1.9067	0.83	174000	1.7662
1.9005	0.84	176000	1.7428
1.8997	0.85	178000	1.7500
1.8963	0.86	180000	1.7297
1.8938	0.87	182000	1.7356
1.8923	0.88	184000	1.7602
1.8896	0.89	186000	1.7426
1.8866	0.9	188000	1.7323
1.887	0.91	190000	1.7587
1.8855	0.92	192000	1.7591
1.8842	0.93	194000	1.7570
1.8808	0.94	196000	1.7311
1.8836	0.95	198000	1.7449
1.8761	0.96	200000	1.7534
1.8721	0.96	202000	1.7623
1.8765	0.97	204000	1.7462
1.8747	0.98	206000	1.7452
1.8667	0.99	208000	1.7303
1.8618	1.0	210000	1.7468
1.8475	1.01	212000	1.7443
1.8435	1.02	214000	1.7622
1.8452	1.03	216000	1.7153
1.84	1.04	218000	1.6976
1.8432	1.05	220000	1.7013
1.842	1.06	222000	1.7073
1.8428	1.07	224000	1.6991
1.841	1.08	226000	1.7477
1.8321	1.09	228000	1.7438
1.838	1.1	230000	1.7352
1.8339	1.11	232000	1.7242
1.836	1.12	234000	1.7221
1.8329	1.13	236000	1.7402
1.8337	1.14	238000	1.7083
1.8267	1.15	240000	1.7200
1.8335	1.16	242000	1.7092
1.8306	1.17	244000	1.7340
1.8279	1.17	246000	1.6983
1.8261	1.18	248000	1.6928
1.8295	1.19	250000	1.7135
1.8227	1.2	252000	1.7156
1.822	1.21	254000	1.7018
1.8216	1.22	256000	1.7157
1.8205	1.23	258000	1.7047
1.8163	1.24	260000	1.6988
1.8187	1.25	262000	1.7077
1.8188	1.26	264000	1.6859
1.8138	1.27	266000	1.6831
1.8173	1.28	268000	1.6887
1.813	1.29	270000	1.6967
1.8114	1.3	272000	1.7085
1.8057	1.31	274000	1.6885
1.8094	1.32	276000	1.7198
1.8079	1.33	278000	1.7036
1.8056	1.34	280000	1.7106
1.8044	1.35	282000	1.6704
1.8047	1.36	284000	1.6811
1.7978	1.37	286000	1.6848
1.7997	1.38	288000	1.6698
1.7997	1.38	290000	1.6820
1.7945	1.39	292000	1.6963
1.7958	1.4	294000	1.6922
1.7923	1.41	296000	1.6577
1.7975	1.42	298000	1.6621
1.7914	1.43	300000	1.6804
1.7944	1.44	302000	1.6953
1.7927	1.45	304000	1.6846
1.789	1.46	306000	1.6889
1.7851	1.47	308000	1.6652
1.7902	1.48	310000	1.6823
1.7873	1.49	312000	1.6603
1.7868	1.5	314000	1.6766
1.7856	1.51	316000	1.6717
1.7807	1.52	318000	1.6466
1.7767	1.53	320000	1.6639
1.7782	1.54	322000	1.6678
1.7762	1.55	324000	1.6853
1.7746	1.56	326000	1.6785
1.7746	1.57	328000	1.6777
1.7716	1.58	330000	1.6784
1.7699	1.59	332000	1.6648
1.7739	1.59	334000	1.6725
1.7703	1.6	336000	1.6915
1.7707	1.61	338000	1.6858
1.7619	1.62	340000	1.6624
1.7652	1.63	342000	1.6797
1.7626	1.64	344000	1.6728
1.7647	1.65	346000	1.6580
1.7616	1.66	348000	1.6679
1.7616	1.67	350000	1.6470
1.7611	1.68	352000	1.6489
1.759	1.69	354000	1.6603
1.7604	1.7	356000	1.6532
1.7599	1.71	358000	1.6477
1.7529	1.72	360000	1.6322
1.7596	1.73	362000	1.6447
1.7508	1.74	364000	1.6509
1.7533	1.75	366000	1.6465
1.755	1.76	368000	1.6485
1.7473	1.77	370000	1.6493
1.7435	1.78	372000	1.6542
1.7483	1.79	374000	1.6573
1.7475	1.8	376000	1.6626
1.7439	1.8	378000	1.6366
1.7417	1.81	380000	1.6312
1.7387	1.82	382000	1.6424
1.7415	1.83	384000	1.6468
1.7409	1.84	386000	1.6528
1.7362	1.85	388000	1.6394
1.7372	1.86	390000	1.6581
1.7347	1.87	392000	1.6546
1.7368	1.88	394000	1.6468
1.7302	1.89	396000	1.6450
1.7317	1.9	398000	1.6368
1.7306	1.91	400000	1.6399
1.7304	1.92	402000	1.6180
1.726	1.93	404000	1.6212
1.7271	1.94	406000	1.6302
1.7312	1.95	408000	1.6264
1.7249	1.96	410000	1.6584
1.7226	1.97	412000	1.6514
1.7214	1.98	414000	1.6516
1.7228	1.99	416000	1.6346
1.7205	2.0	418000	1.6370
1.7041	2.01	420000	1.6021
1.691	2.02	422000	1.6385
1.6896	2.02	424000	1.6280
1.6882	2.03	426000	1.6295
1.6889	2.04	428000	1.6445
1.6904	2.05	430000	1.6558
1.6933	2.06	432000	1.6164
1.6916	2.07	434000	1.6011
1.6873	2.08	436000	1.6199
1.6903	2.09	438000	1.6300
1.6859	2.1	440000	1.6104
1.6901	2.11	442000	1.6248
1.6884	2.12	444000	1.6251
1.6859	2.13	446000	1.6145
1.6906	2.14	448000	1.6181
1.6859	2.15	450000	1.6264
1.6814	2.16	452000	1.6069
1.6853	2.17	454000	1.6089
1.6881	2.18	456000	1.6102
1.6869	2.19	458000	1.6327
1.6827	2.2	460000	1.6069
1.6813	2.21	462000	1.6278
1.6806	2.22	464000	1.6176
1.6763	2.23	466000	1.6180
1.68	2.23	468000	1.6226
1.6816	2.24	470000	1.6071
1.6845	2.25	472000	1.6178
1.6764	2.26	474000	1.6073
1.682	2.27	476000	1.5966
1.6727	2.28	478000	1.5979
1.6718	2.29	480000	1.6109
1.6764	2.3	482000	1.6034
1.671	2.31	484000	1.6001
1.6691	2.32	486000	1.6148
1.6706	2.33	488000	1.6003
1.6705	2.34	490000	1.6021
1.6699	2.35	492000	1.5940
1.6708	2.36	494000	1.6077
1.6715	2.37	496000	1.6188
1.6672	2.38	498000	1.5903
1.6638	2.39	500000	1.6042
1.6634	2.4	502000	1.5967
1.6669	2.41	504000	1.5904
1.6643	2.42	506000	1.6071
1.6606	2.43	508000	1.6065
1.6573	2.44	510000	1.6010
1.6603	2.44	512000	1.5801
1.6568	2.45	514000	1.5961
1.6564	2.46	516000	1.6020
1.6596	2.47	518000	1.5952
1.6567	2.48	520000	1.5760
1.6536	2.49	522000	1.5697
1.6564	2.5	524000	1.5664
1.652	2.51	526000	1.5616
1.653	2.52	528000	1.5738
1.6525	2.53	530000	1.5754
1.65	2.54	532000	1.5749
1.6519	2.55	534000	1.5788
1.6515	2.56	536000	1.5953
1.6492	2.57	538000	1.5836
1.6473	2.58	540000	1.5896
1.6452	2.59	542000	1.5858
1.6464	2.6	544000	1.5760
1.6445	2.61	546000	1.5683
1.6457	2.62	548000	1.5823
1.6417	2.63	550000	1.5780
1.6407	2.64	552000	1.5715
1.6368	2.65	554000	1.5618
1.6357	2.65	556000	1.5725
1.6446	2.66	558000	1.5744
1.634	2.67	560000	1.5360
1.6351	2.68	562000	1.5599
1.6362	2.69	564000	1.5607
1.637	2.7	566000	1.5561
1.6324	2.71	568000	1.5591
1.6325	2.72	570000	1.5527
1.6323	2.73	572000	1.5537
1.629	2.74	574000	1.5673
1.627	2.75	576000	1.5509
1.6279	2.76	578000	1.5507
1.6291	2.77	580000	1.5304
1.625	2.78	582000	1.5540
1.6246	2.79	584000	1.5530
1.6228	2.8	586000	1.5570
1.6241	2.81	588000	1.5586
1.6224	2.82	590000	1.5480
1.6264	2.83	592000	1.5624
1.6214	2.84	594000	1.5565
1.6187	2.85	596000	1.5397
1.6191	2.86	598000	1.5520
1.6192	2.87	600000	1.5494
1.6182	2.87	602000	1.5608
1.6164	2.88	604000	1.5428
1.6107	2.89	606000	1.5525
1.614	2.9	608000	1.5277
1.6158	2.91	610000	1.5502
1.6082	2.92	612000	1.5452
1.6089	2.93	614000	1.5400
1.6112	2.94	616000	1.5322
1.6069	2.95	618000	1.5394
1.6111	2.96	620000	1.5537
1.6038	2.97	622000	1.5486
1.6073	2.98	624000	1.5551
1.6046	2.99	626000	1.5386
1.6051	3.0	628000	1.5369
1.5672	3.01	630000	1.5361
1.5694	3.02	632000	1.5390
1.5692	3.03	634000	1.5386
1.5651	3.04	636000	1.5456
1.5724	3.05	638000	1.5419
1.5708	3.06	640000	1.5363
1.5665	3.07	642000	1.5446
1.5706	3.08	644000	1.5331
1.5679	3.08	646000	1.5449
1.5678	3.09	648000	1.5436
1.5676	3.1	650000	1.5309
1.5657	3.11	652000	1.5334
1.5697	3.12	654000	1.5303
1.5617	3.13	656000	1.5380
1.5675	3.14	658000	1.5404
1.5612	3.15	660000	1.5258
1.5639	3.16	662000	1.5329
1.567	3.17	664000	1.5418
1.5619	3.18	666000	1.5314
1.5637	3.19	668000	1.5201
1.5608	3.2	670000	1.5181
1.5641	3.21	672000	1.5290
1.5626	3.22	674000	1.5180
1.5605	3.23	676000	1.5156
1.5566	3.24	678000	1.5266
1.5587	3.25	680000	1.5286
1.5602	3.26	682000	1.5265
1.5535	3.27	684000	1.5354
1.5589	3.28	686000	1.5265
1.5569	3.29	688000	1.5346
1.559	3.29	690000	1.5306
1.5507	3.3	692000	1.5359
1.5547	3.31	694000	1.5264
1.5498	3.32	696000	1.5264
1.5559	3.33	698000	1.5273
1.553	3.34	700000	1.5137
1.5503	3.35	702000	1.5143
1.5498	3.36	704000	1.5263
1.5516	3.37	706000	1.5096
1.5461	3.38	708000	1.5112
1.5489	3.39	710000	1.5094
1.5451	3.4	712000	1.5079
1.544	3.41	714000	1.5058
1.5446	3.42	716000	1.5005
1.5417	3.43	718000	1.4972
1.5469	3.44	720000	1.5043
1.5407	3.45	722000	1.5041
1.5484	3.46	724000	1.5104
1.5409	3.47	726000	1.5087
1.5431	3.48	728000	1.5114
1.5393	3.49	730000	1.5102
1.5364	3.5	732000	1.5143
1.5403	3.5	734000	1.5202
1.5386	3.51	736000	1.5143
1.5381	3.52	738000	1.5198
1.5341	3.53	740000	1.5136
1.5344	3.54	742000	1.5172
1.5347	3.55	744000	1.5149
1.5292	3.56	746000	1.5141
1.5344	3.57	748000	1.5066
1.5307	3.58	750000	1.5087
1.5324	3.59	752000	1.5113
1.5273	3.6	754000	1.5101
1.5273	3.61	756000	1.4975
1.5282	3.62	758000	1.5053
1.5252	3.63	760000	1.4998
1.525	3.64	762000	1.5020
1.5297	3.65	764000	1.5075
1.5215	3.66	766000	1.4980
1.5237	3.67	768000	1.5066
1.5248	3.68	770000	1.5093
1.5231	3.69	772000	1.5090
1.5224	3.7	774000	1.5093
1.526	3.71	776000	1.5015
1.5215	3.71	778000	1.5045
1.5231	3.72	780000	1.4971
1.5205	3.73	782000	1.4987
1.5171	3.74	784000	1.5001
1.5134	3.75	786000	1.4951
1.5155	3.76	788000	1.4975
1.5154	3.77	790000	1.4928
1.5167	3.78	792000	1.4983
1.5146	3.79	794000	1.4938
1.5138	3.8	796000	1.4985
1.5137	3.81	798000	1.5021
1.5111	3.82	800000	1.5020
1.5134	3.83	802000	1.4998
1.5086	3.84	804000	1.5001
1.5081	3.85	806000	1.5031
1.5097	3.86	808000	1.5008
1.5128	3.87	810000	1.4990
1.5093	3.88	812000	1.4994
1.5109	3.89	814000	1.5021
1.5049	3.9	816000	1.5012
1.5042	3.91	818000	1.5013
1.5053	3.92	820000	1.4946
1.5066	3.93	822000	1.4984
1.5074	3.93	824000	1.4963
1.5046	3.94	826000	1.4972
1.5043	3.95	828000	1.4970
1.5064	3.96	830000	1.4940
1.4999	3.97	832000	1.4940
1.5022	3.98	834000	1.4934
1.5054	3.99	836000	1.4929

Framework versions

• Transformers 4.19.2
• Pytorch 1.11.0+cu113
• Datasets 2.2.2
• Tokenizers 0.12.1