nllb_complete

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

• Loss: 1.8285
• Bleu: 17.1412
• Gen Len: 17.896

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: 3e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 5000
• num_epochs: 24.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
2.1296	1.4834	10000	2.0709	9.9056	20.1323
2.0253	2.9668	20000	1.9697	11.7423	19.27
1.8771	4.4503	30000	1.9199	13.3983	18.9643
1.7891	5.9338	40000	1.8851	14.1016	18.3833
1.7159	7.4173	50000	1.8680	14.8584	18.2797
1.6594	8.9007	60000	1.8473	15.8809	18.3863
1.6609	10.3842	70000	1.8406	15.8588	18.159
1.6358	11.8676	80000	1.8319	16.4395	18.4773
1.5623	13.3511	90000	1.8298	16.8956	18.3217
1.5534	14.8345	100000	1.8218	16.8725	18.5327
1.498	16.3180	110000	1.8286	16.6418	17.9697
1.4663	17.8014	120000	1.8252	17.2847	17.9357
1.4309	19.2849	130000	1.8299	17.027	17.7263
1.4398	20.7684	140000	1.8270	17.0189	18.1353
1.4534	22.2519	150000	1.8292	17.04	17.9637
1.4441	23.7353	160000	1.8285	17.1412	17.896

Framework versions

• Transformers 4.50.3
• Pytorch 2.7.0+cu126
• Datasets 3.5.0
• Tokenizers 0.21.1