opus-ecolindo

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

• Loss: 1.1254
• Bleu: 35.4242

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: 1e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 4000
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu
1.4457	1.0	21019	1.1641	35.6285
1.1877	2.0	42038	1.1141	35.9198
1.1049	3.0	63057	1.0936	36.0605
1.049	4.0	84076	1.0821	36.0444
1.0051	5.0	105095	1.0794	36.6179
0.9685	6.0	126114	1.0751	36.7172
0.9376	7.0	147133	1.0751	36.407
0.9098	8.0	168152	1.0759	36.7071
0.8859	9.0	189171	1.0809	36.0112
0.864	10.0	210190	1.0800	36.1705
0.844	11.0	231209	1.0861	36.1114
0.8261	12.0	252228	1.0862	35.9952
0.8102	13.0	273247	1.0944	35.659
0.7945	14.0	294266	1.0982	35.5881
0.7812	15.0	315285	1.1019	35.8675
0.7688	16.0	336304	1.1066	35.5418
0.7576	17.0	357323	1.1087	35.7305
0.7472	18.0	378342	1.1116	35.4516
0.7377	19.0	399361	1.1162	35.3253
0.7298	20.0	420380	1.1185	35.2614
0.7233	21.0	441399	1.1203	35.5355
0.7162	22.0	462418	1.1220	35.3226
0.7116	23.0	483437	1.1226	35.3893
0.7075	24.0	504456	1.1250	35.4201
0.7037	25.0	525475	1.1254	35.4242

Framework versions

• Transformers 4.26.1
• Pytorch 2.0.0
• Datasets 2.10.1
• Tokenizers 0.11.0