Mizomade-Mizo-EN-NLLB-S1

About Model

Mizomade-Mizo-EN-S1 is a fine-tuned NLLB-200 distilled 600M neural machine translation model developed for Mizo-English translation. The model is based on Meta AI's facebook/nllb-200-distilled-600M model and was fine-tuned on Mizo-English parallel translation data to improve translation quality for the Mizo language, a low-resource language. The model supports Mizo-English and English-Mizo machine translation and is intended primarily for research, evaluation, and development of Mizo language technologies. Model Architecture: NLLB-200 / M2M100ForConditionalGeneration Base Model: facebook/nllb-200-distilled-600M Task: Machine Translation Language Pair: Mizo ↔ English Model Size: +600M parameters Fine-tuning: Supervised fine-tuning on Mizo-English parallel data License: CC-BY-NC-4.0 Training hyperparameters The following hyperparameters were used during training: learning_rate: 2e-05 train_batch_size: 4 eval_batch_size: 8 seed: 42 gradient_accumulation_steps: 8 total_train_batch_size: 32 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: cosine num_epochs: 3 label_smoothing_factor: 0.1