enko_mbartLarge_36p_run1

This model is a fine-tuned version of facebook/mbart-large-50-many-to-many-mmt on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1249
• Bleu: 26.0278
• Gen Len: 16.4716

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• total_eval_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
1.3157	0.46	5000	1.2895	21.6734	16.4931
1.2575	0.93	10000	1.2279	22.5814	16.8009
1.1578	1.39	15000	1.1733	24.2386	16.5838
1.0885	1.86	20000	1.1464	24.7921	16.6649
1.0451	2.32	25000	1.1437	24.8186	16.5188
1.0465	2.79	30000	1.1425	24.9684	16.4987
1.0156	3.25	35000	1.1464	25.0127	16.5515
0.9893	3.72	40000	1.1544	24.6323	16.6096
0.8779	4.18	45000	1.1419	25.3982	16.457
0.8565	4.65	50000	1.1249	26.0152	16.4749
0.7293	5.11	55000	1.1566	25.8209	16.3462
0.7294	5.57	60000	1.1824	25.061	16.3295
0.7254	6.04	65000	1.2153	24.8712	16.381
0.6469	6.5	70000	1.1965	25.3326	16.5447

Framework versions

• Transformers 4.34.1
• Pytorch 2.1.0+cu121
• Datasets 2.14.6
• Tokenizers 0.14.1