ja2en_helsinki

This model is a fine-tuned version of Helsinki-NLP/opus-mt-ja-en on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 1.4235
  • Bleu: 24.3293
  • Gen Len: 34.9345

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: 2e-05
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 10
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Bleu Gen Len
1.8739 1.0 27518 1.7275 20.8396 34.3259
1.6828 2.0 55036 1.5887 21.5972 34.4103
1.5819 3.0 82554 1.5258 22.7249 34.5991
1.4986 4.0 110072 1.4865 23.262 34.7543
1.4142 5.0 137590 1.4625 23.2547 34.8388
1.3783 6.0 165108 1.4501 23.7076 34.4905
1.333 7.0 192626 1.4357 23.7245 34.7474
1.3132 8.0 220144 1.4281 23.9473 35.006
1.2922 9.0 247662 1.4249 24.0284 34.9741
1.2698 10.0 275180 1.4235 24.3293 34.9345

Framework versions

  • Transformers 4.45.2
  • Pytorch 2.5.0+cu118
  • Datasets 3.0.1
  • Tokenizers 0.20.1
Downloads last month
96
Safetensors
Model size
75.3M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for kknssro/ja2en_helsinki

Finetuned
(3)
this model