enko_mbartLarge_36p_tokenize_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: 38.8566
• 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	34.4176	16.4931
1.2575	0.93	10000	1.2279	35.0029	16.8009
1.1578	1.39	15000	1.1733	36.9282	16.5838
1.0885	1.86	20000	1.1464	37.6913	16.6649
1.0451	2.32	25000	1.1437	37.7875	16.5188
1.0465	2.79	30000	1.1425	37.895	16.4987
1.0156	3.25	35000	1.1464	37.8434	16.5515
0.9893	3.72	40000	1.1544	37.358	16.6096
0.8779	4.18	45000	1.1419	38.1772	16.457
0.8565	4.65	50000	1.1249	38.8455	16.4749
0.7293	5.11	55000	1.1566	38.4853	16.3462
0.7294	5.57	60000	1.1824	37.8822	16.3295
0.7254	6.04	65000	1.2153	37.3612	16.381

Framework versions

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