ind-to-bbc-nmt-v6

This model is a fine-tuned version of facebook/nllb-200-distilled-600M on the nusatranslation_mt dataset. It achieves the following results on the evaluation set:

• Loss: 1.1587
• Sacrebleu: 31.0331
• Gen Len: 45.1815

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Sacrebleu	Gen Len
4.3415	1.0	825	1.6110	27.0363	45.267
1.415	2.0	1650	1.2550	30.5956	45.5
1.1044	3.0	2475	1.1769	31.2342	45.4315
0.951	4.0	3300	1.1532	31.8633	45.149
0.8409	5.0	4125	1.1340	31.5171	45.355
0.7582	6.0	4950	1.1273	31.0686	45.222
0.6937	7.0	5775	1.1387	31.3129	45.1355
0.6433	8.0	6600	1.1479	31.444	45.233
0.6056	9.0	7425	1.1521	31.3122	45.0945
0.5819	10.0	8250	1.1587	31.0331	45.1815

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.14.6
• Tokenizers 0.19.1