english_to_taglish_8483_v2

This model is a fine-tuned version of facebook/nllb-200-distilled-1.3B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.0151

Model description

Fixed structure and corrected some data in training.

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: 1e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• 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_with_restarts
• lr_scheduler_warmup_steps: 750
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss
1.7899	1.0	54	1.6850
1.7335	2.0	108	1.5802
1.593	3.0	162	1.3848
1.3898	4.0	216	1.2504
1.2958	5.0	270	1.2021
1.258	6.0	324	1.1684
1.2047	7.0	378	1.1390
1.1975	8.0	432	1.1171
1.1593	9.0	486	1.0993
1.1221	10.0	540	1.0846
1.1018	11.0	594	1.0704
1.0643	12.0	648	1.0574
1.0284	13.0	702	1.0463
1.0013	14.0	756	1.0364
0.9785	15.0	810	1.0276
0.9346	16.0	864	1.0203
0.9385	17.0	918	1.0184
0.92	18.0	972	1.0168
0.9074	19.0	1026	1.0151
0.8907	20.0	1080	1.0151

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.1