opus-mt-finetuned-su-to-id

This model is a fine-tuned version of hermanshid/opus-mt-finetuned-id-to-su on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.0114
• Bleu: 34.1024
• Gen Len: 54.7788

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
1.4771	1.0	2500	1.3476	26.9951	54.326
1.2773	2.0	5000	1.2004	29.6076	54.7672
1.1813	3.0	7500	1.1286	31.2619	55.0676
1.1052	4.0	10000	1.0859	32.1006	54.8726
1.0558	5.0	12500	1.0561	32.7946	54.7328
1.019	6.0	15000	1.0383	33.2759	54.7428
0.9953	7.0	17500	1.0280	33.5529	54.6836
0.9656	8.0	20000	1.0169	33.9846	54.8196
0.9456	9.0	22500	1.0132	34.013	54.8898
0.9417	10.0	25000	1.0114	34.1024	54.7788

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3