Edit model card

LN_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: 2.5382

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: 64
  • eval_batch_size: 64
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 3
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss
2.5108 0.02 2000 2.8405
2.2729 0.04 4000 2.7758
2.1673 0.06 6000 2.7098
2.0981 0.08 8000 2.6487
2.048 0.1 10000 2.7008
2.0077 0.12 12000 2.6614
1.9677 0.13 14000 2.6227
1.9445 0.15 16000 2.5895
1.9137 0.17 18000 2.5897
1.8911 0.19 20000 2.6771
1.8695 0.21 22000 2.6332
1.8479 0.23 24000 2.6130
1.8378 0.25 26000 2.6518
1.8191 0.27 28000 2.6401
1.8024 0.29 30000 2.6617
1.7933 0.31 32000 2.6705
1.7787 0.33 34000 2.6280
1.7661 0.35 36000 2.6911
1.7514 0.36 38000 2.6766
1.7444 0.38 40000 2.6996
1.7363 0.4 42000 2.6276
1.722 0.42 44000 2.6466
1.7177 0.44 46000 2.5937
1.7055 0.46 48000 2.6386
1.6956 0.48 50000 2.6794
1.6885 0.5 52000 2.7336
1.679 0.52 54000 2.7266
1.6715 0.54 56000 2.6945
1.6666 0.56 58000 2.7111
1.6599 0.58 60000 2.7205
1.6566 0.59 62000 2.7194
1.6481 0.61 64000 2.6582
1.6434 0.63 66000 2.6859
1.6315 0.65 68000 2.7058
1.6258 0.67 70000 2.7428
1.6189 0.69 72000 2.7411
1.6169 0.71 74000 2.7039
1.6087 0.73 76000 2.6844
1.6021 0.75 78000 2.6454
1.6034 0.77 80000 2.6596
1.5941 0.79 82000 2.6903
1.5862 0.81 84000 2.7099
1.5836 0.83 86000 2.6929
1.5827 0.84 88000 2.7181
1.5747 0.86 90000 2.6888
1.5678 0.88 92000 2.6662
1.5643 0.9 94000 2.6663
1.561 0.92 96000 2.6699
1.5565 0.94 98000 2.6667
1.5501 0.96 100000 2.6828
1.5476 0.98 102000 2.6531
1.5444 1.0 104000 2.6799
1.5057 1.02 106000 2.6525
1.5003 1.04 108000 2.6996
1.4996 1.06 110000 2.6649
1.4996 1.07 112000 2.6974
1.4966 1.09 114000 2.7594
1.4967 1.11 116000 2.6966
1.492 1.13 118000 2.6929
1.4923 1.15 120000 2.6522
1.4838 1.17 122000 2.6363
1.4839 1.19 124000 2.6849
1.4807 1.21 126000 2.6667
1.4778 1.23 128000 2.6684
1.4731 1.25 130000 2.6338
1.4727 1.27 132000 2.6093
1.4695 1.29 134000 2.6020
1.4656 1.3 136000 2.6341
1.4648 1.32 138000 2.6509
1.4578 1.34 140000 2.6807
1.4606 1.36 142000 2.6357
1.4529 1.38 144000 2.6404
1.4488 1.4 146000 2.6347
1.4442 1.42 148000 2.6058
1.4447 1.44 150000 2.6645
1.4432 1.46 152000 2.6070
1.437 1.48 154000 2.5987
1.4345 1.5 156000 2.6309
1.43 1.52 158000 2.5947
1.4301 1.54 160000 2.5938
1.4267 1.55 162000 2.6146
1.426 1.57 164000 2.6519
1.4193 1.59 166000 2.6163
1.416 1.61 168000 2.5793
1.4146 1.63 170000 2.6031
1.4091 1.65 172000 2.5826
1.4067 1.67 174000 2.5891
1.4081 1.69 176000 2.6006
1.4023 1.71 178000 2.5697
1.4003 1.73 180000 2.5633
1.3986 1.75 182000 2.5494
1.3924 1.77 184000 2.5577
1.3931 1.78 186000 2.5888
1.3851 1.8 188000 2.5716
1.3869 1.82 190000 2.5570
1.3825 1.84 192000 2.5702
1.3787 1.86 194000 2.5754
1.3738 1.88 196000 2.5901
1.3734 1.9 198000 2.5374
1.3693 1.92 200000 2.5897
1.3703 1.94 202000 2.5422
1.3685 1.96 204000 2.5825
1.3664 1.98 206000 2.5201
1.3607 2.0 208000 2.5733
1.3217 2.02 210000 2.5879
1.31 2.03 212000 2.5777
1.3125 2.05 214000 2.5724
1.3084 2.07 216000 2.5968
1.3087 2.09 218000 2.5976
1.3063 2.11 220000 2.5969
1.3057 2.13 222000 2.6353
1.3067 2.15 224000 2.6147
1.3013 2.17 226000 2.5897
1.3018 2.19 228000 2.5783
1.2968 2.21 230000 2.6172
1.2975 2.23 232000 2.6180
1.2946 2.25 234000 2.6192
1.299 2.26 236000 2.5895
1.2896 2.28 238000 2.5682
1.287 2.3 240000 2.5653
1.2902 2.32 242000 2.5501
1.2862 2.34 244000 2.5747
1.2841 2.36 246000 2.5654
1.2838 2.38 248000 2.5703
1.2813 2.4 250000 2.5919
1.2778 2.42 252000 2.5552
1.2821 2.44 254000 2.5603
1.2729 2.46 256000 2.5455
1.2718 2.48 258000 2.5688
1.2729 2.49 260000 2.5574
1.2699 2.51 262000 2.5468
1.2677 2.53 264000 2.5704
1.2647 2.55 266000 2.5665
1.2628 2.57 268000 2.5594
1.2636 2.59 270000 2.5426
1.2573 2.61 272000 2.5666
1.2576 2.63 274000 2.5580
1.2511 2.65 276000 2.5742
1.2513 2.67 278000 2.5646
1.2495 2.69 280000 2.5669
1.2472 2.71 282000 2.5700
1.2478 2.73 284000 2.5496
1.2471 2.74 286000 2.5335
1.2436 2.76 288000 2.5315
1.2411 2.78 290000 2.5302
1.2391 2.8 292000 2.5290
1.2352 2.82 294000 2.5303
1.2332 2.84 296000 2.5412
1.233 2.86 298000 2.5523
1.2298 2.88 300000 2.5524
1.2285 2.9 302000 2.5517
1.2297 2.92 304000 2.5419
1.2256 2.94 306000 2.5404
1.2239 2.96 308000 2.5390
1.2264 2.97 310000 2.5364
1.2259 2.99 312000 2.5382

Framework versions

  • Transformers 4.19.2
  • Pytorch 1.11.0+cu113
  • Datasets 2.2.2
  • Tokenizers 0.12.1
Downloads last month
5