opus-mt-finetuned-id-to-jv

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

• Loss: 0.5597
• Bleu: 50.74
• Gen Len: 58.1428

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
0.841	1.0	2500	0.7481	44.6803	58.0716
0.7025	2.0	5000	0.6599	47.0415	58.3842
0.6305	3.0	7500	0.6203	48.4781	58.154
0.5772	4.0	10000	0.5969	49.1335	58.4164
0.5472	5.0	12500	0.5816	49.7317	58.149
0.5215	6.0	15000	0.5728	50.1163	58.0292
0.5079	7.0	17500	0.5676	50.4371	58.2302
0.4845	8.0	20000	0.5626	50.606	58.0254
0.4703	9.0	22500	0.5600	50.7025	58.0016
0.4597	10.0	25000	0.5597	50.74	58.1428

Framework versions

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