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: 0.5537
• Bleu: 54.0987
• Gen Len: 17.1547

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 32
• 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
0.7527	2.0916	10000	0.7154	41.0087	17.0123
0.5274	4.1832	20000	0.5901	46.2079	17.2597
0.4209	6.2748	30000	0.5502	49.6827	17.0787
0.381	8.3665	40000	0.5324	51.0468	17.1997
0.3123	10.4581	50000	0.5264	52.239	17.0687
0.279	12.5497	60000	0.5292	52.8077	17.163
0.2568	14.6413	70000	0.5320	53.2148	17.1983
0.2234	16.7329	80000	0.5415	53.2988	17.1817
0.2208	18.8245	90000	0.5455	53.9008	17.1253
0.2179	20.9162	100000	0.5501	54.2302	17.107
0.2057	23.0077	110000	0.5537	54.0987	17.1547

Framework versions

• Transformers 4.53.3
• Pytorch 2.7.1+cu126
• Datasets 3.6.0
• Tokenizers 0.21.2