mbart-large-50-finetuned-en-to-ba

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: 2.6372
• Bleu: 6.3802
• Gen Len: 47.071

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 0
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• training_steps: 12000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
2.1365	0.3476	1000	2.7681	2.9925	52.131
1.4326	0.6952	2000	2.5934	4.1353	45.036
1.2369	1.0428	3000	2.5647	5.0215	49.007
1.0832	1.3903	4000	2.5558	5.0661	48.543
1.036	1.7379	5000	2.5203	5.4399	47.738
0.9671	2.0855	6000	2.5657	5.8593	49.639
0.8711	2.4331	7000	2.5162	5.6784	47.706
0.8561	2.7807	8000	2.5742	5.9194	47.395
0.8103	3.1283	9000	2.6108	6.3266	47.508
0.7534	3.4758	10000	2.6139	6.22	48.42
0.747	3.8234	11000	2.6008	6.3408	46.571
0.7123	4.1710	12000	2.6372	6.3802	47.071

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0