Bamalingua-10

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: 1.2686
• Bleu: 33.3172

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 500
• num_epochs: 9

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu
1.6029	1.0	1580	1.3671	19.8676
1.133	2.0	3160	1.2158	23.8416
0.83	3.0	4740	1.1721	25.9122
0.6113	4.0	6320	1.1634	28.6661
0.4468	5.0	7900	1.1735	31.1171
0.3295	6.0	9480	1.2041	32.4465
0.2543	7.0	11060	1.2353	33.0069
0.2104	8.0	12640	1.2586	33.3333
0.187	9.0	14220	1.2686	33.3172

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu126
• Datasets 4.4.2
• Tokenizers 0.22.1