opus-mt-id-en-open-subtitles

This model was trained from scratch on the open_subtitles dataset. It achieves the following results on the evaluation set:

• Loss: 1.8430
• Bleu: 36.9382

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: 0.0001
• 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
1.3533	1.0	28125	1.3274	37.6662
1.2814	2.0	56250	1.3525	37.5909
1.2058	3.0	84375	1.3674	37.8008
1.1415	4.0	112500	1.3722	37.4849
1.0842	5.0	140625	1.3943	37.7558
1.0309	6.0	168750	1.3994	37.6332
0.9802	7.0	196875	1.4216	37.7529
0.9316	8.0	225000	1.4304	37.9906
0.8838	9.0	253125	1.4462	37.7833
0.8378	10.0	281250	1.4639	37.5971
0.7921	11.0	309375	1.4859	37.6285
0.7484	12.0	337500	1.5060	37.5413
0.7043	13.0	365625	1.5256	37.5118
0.6622	14.0	393750	1.5555	37.5092
0.6208	15.0	421875	1.5733	37.2924
0.5807	16.0	450000	1.6048	37.319
0.542	17.0	478125	1.6435	37.0629
0.5043	18.0	506250	1.6647	37.1334
0.4685	19.0	534375	1.7014	37.02
0.4352	20.0	562500	1.7300	36.9514
0.4031	21.0	590625	1.7572	36.9637
0.3731	22.0	618750	1.7902	36.9821
0.346	23.0	646875	1.8112	36.9586
0.3227	24.0	675000	1.8325	36.9286
0.303	25.0	703125	1.8430	36.9382

Framework versions

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