nllb-sa-finetuned

This model is a fine-tuned version of facebook/nllb-200-3.3B on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1843
• Model Preparation Time: 0.0575

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: 0.0003
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_BNB 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
• training_steps: 10000

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time
3.1615	0.0101	500	1.4611	0.0575
3.0072	0.0202	1000	1.3972	0.0575
2.8503	0.0303	1500	1.3589	0.0575
2.8210	0.0404	2000	1.3341	0.0575
2.8089	0.0505	2500	1.3136	0.0575
2.7953	0.0606	3000	1.2988	0.0575
2.7414	0.0707	3500	1.2842	0.0575
2.6644	0.0808	4000	1.2661	0.0575
2.6363	0.0909	4500	1.2535	0.0575
2.6871	0.1010	5000	1.2412	0.0575
2.5607	0.1111	5500	1.2289	0.0575
2.6463	0.1212	6000	1.2177	0.0575
2.6042	0.1313	6500	1.2101	0.0575
2.5593	0.1414	7000	1.2027	0.0575
2.6148	0.1515	7500	1.1949	0.0575
2.5434	0.1616	8000	1.1908	0.0575
2.5353	0.1717	8500	1.1868	0.0575
2.5028	0.1818	9000	1.1856	0.0575
2.5299	0.1919	9500	1.1844	0.0575
2.4559	0.2020	10000	1.1843	0.0575

Framework versions

• PEFT 0.19.1
• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2