meta-nllb-600m-mt-en-twi-v4

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: 0.5793
• Rouge1: 0.6092
• Bleu: 22.4778

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: 1e-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
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Bleu
No log	1.0	480	4.1464	0.5420	15.3433
5.9598	2.0	960	1.8878	0.5603	16.8733
2.993	3.0	1440	0.7451	0.5753	18.6067
1.3619	4.0	1920	0.6291	0.5880	19.9709
0.888	5.0	2400	0.6059	0.5953	20.4567
0.7774	6.0	2880	0.5961	0.6000	21.1082
0.7358	7.0	3360	0.5907	0.6049	21.4798
0.6934	8.0	3840	0.5866	0.6068	21.6956
0.666	9.0	4320	0.5816	0.6058	21.8561
0.6533	10.0	4800	0.5799	0.6063	21.8737
0.6266	11.0	5280	0.5791	0.6078	22.1400
0.6063	12.0	5760	0.5792	0.6106	22.3387
0.6058	13.0	6240	0.5790	0.6072	22.2070
0.5786	14.0	6720	0.5777	0.6084	22.2723
0.5754	15.0	7200	0.5800	0.6079	22.2117
0.5707	16.0	7680	0.5784	0.6084	22.2791
0.557	17.0	8160	0.5790	0.6081	22.4436
0.5531	18.0	8640	0.5796	0.6097	22.5290
0.5464	19.0	9120	0.5797	0.6085	22.3927
0.5499	20.0	9600	0.5793	0.6092	22.4778

Framework versions

• Transformers 4.44.2
• Pytorch 2.3.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1