Beta
Explore
Marketplace
Neural Labs
Chat
Wallet
Docs
za-zh-rs-za – AI Model by FiveC | AlphaNeural AI
You can deploy this model and start earning money today!
FiveC
/
za-zh-rs-za
like
0
transformers
safetensors
mbart
text2text-generation
generated_from_trainer
facebook/mbart-large-50-many-to-many-mmt
finetune
endpoints_compatible
us
Views
No views yet
Model card
Files and Versions
Community
API
Deploy
za-zh-rs-za
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: 3.2987
Sacrebleu: 2.6285
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: 16
eval_batch_size: 16
seed: 42
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
num_epochs: 3
mixed_precision_training: Native AMP
Training results
Training Loss
Epoch
Step
Validation Loss
Sacrebleu
1.9199
1.0
1236
3.2717
1.9790
1.1039
2.0
2472
3.2611
0.9655
0.7351
3.0
3708
3.2987
2.6285
Framework versions
Transformers 5.0.0
Pytorch 2.10.0+cu128
Datasets 4.0.0
Tokenizers 0.22.2
