test_llm_nllb_100_e_12_lr3e5_ada

This model is a fine-tuned version of facebook/nllb-200-distilled-600M on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5388
• Rouge1: 0.6155
• Rouge2: 0.3817
• Rougel: 0.57
• Sacrebleu: 23.323

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: 237
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 12
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Sacrebleu
0.5121	1.0	2040	0.5072	0.5877	0.3511	0.5432	20.8262
0.4381	2.0	4080	0.4802	0.6022	0.3684	0.5572	21.9614
0.3465	3.0	6120	0.4816	0.6143	0.3806	0.569	23.0564
0.3084	4.0	8160	0.4832	0.6172	0.3872	0.5726	23.4185
0.2836	5.0	10200	0.4902	0.6196	0.3883	0.5751	23.7412
0.2479	6.0	12240	0.5001	0.6182	0.3832	0.5719	23.1833
0.2036	7.0	14280	0.5112	0.6191	0.3865	0.5746	23.1771
0.1973	8.0	16320	0.5190	0.6207	0.3865	0.5735	23.2226
0.1615	9.0	18360	0.5258	0.619	0.3877	0.5733	23.7005
0.1546	10.0	20400	0.5335	0.6172	0.3835	0.5708	23.458
0.1345	11.0	22440	0.5336	0.6125	0.3786	0.5665	23.1359
0.1294	12.0	24480	0.5388	0.6155	0.3817	0.57	23.323

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0
• Datasets 2.21.0
• Tokenizers 0.19.1