nllb-en-te

This model is a fine-tuned version of facebook/nllb-200-distilled-600M on jaksani/english-to-telugu dataset.

Model description

This model is fine-tuned from facebook/nllb-200-distilled-600M for English-to-Telugu machine translation.

Intended uses & limitations

This model is fine-tuned for English-to-Telugu machine translation. It is intended to translate English sentences into natural Telugu text using the pretrained multilingual capabilities of the NLLB-200 model.

Potential use cases include:

• English-to-Telugu text translation
• Educational and research purposes
• NLP experimentation with machine translation
• Building multilingual applications and translation systems

Training and evaluation data

• The model is fine-tuned only for English-to-Telugu translation.
• Translation quality depends on the quality and coverage of the fine-tuning dataset.
• The model may not perform well on highly technical, legal, medical, or domain-specific text if such examples were underrepresented in the training data.
• The model may generate incorrect or less fluent translations for very long or complex sentences.
• The model has not been evaluated for safety-critical applications.

Training procedure

The model was fine-tuned using the jaksani/english-to-telugu dataset available on the Hugging Face Hub.

• Source Language: English
• Target Language: Telugu
• Task: Neural Machine Translation

Data Split

Since the dataset did not provide predefined validation and test splits, the training dataset was randomly split into:

• Training Set: 95%
• Validation Set: 5%

The validation set was used to monitor model performance during fine-tuning.

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• 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

Framework versions

• Transformers 5.13.1
• Pytorch 2.11.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2