opus-mt-lg-en-finetuned

This model performs translation trained using MLflow and deployed on Hugging Face.

Model Details

• Model Name: opus-mt-lg-en-finetuned
• Version: 6
• Task: Translation
• Languages: lg,en
• Framework: pytorch
• License: apache-2.0

Intended Uses & Limitations

Intended Uses

• Translation tasks
• Research and development
• Child helpline services support

Limitations

• Performance may vary on out-of-distribution data
• Should be evaluated on your specific use case before production deployment
• Designed for child helpline contexts, may need adaptation for other domains

Training Data

• Dataset: luganda_parallel.json
• Size: Not specified
• Languages: lg,en

Training Configuration

Parameter	Value
Dataset Config Custom Datasets	['dataset/custom/luganda/luganda_parallel.jsonl']
Dataset Config Max Samples	None
Dataset Config Primary Dataset	custom
Dataset Config Validation Split	0.15
Evaluation Config Metrics	['bleu', 'chrf', 'meteor']
Evaluation Config Test Size	150
Language Name	Luganda
Language Pair	lg-en
Max Length	512
Model Name	Helsinki-NLP/opus-mt-mul-en
Team Config Assigned Developer	Marlon
Team Config Notes	Increased Batch by 2, epochs by 2 and max length * 2
Team Config Priority	medium
Total Parameters	77518848
Trainable Parameters	76994560
Training Config Batch Size	4
Training Config Learning Rate	2e-05
Training Config Max Length	256
Training Config Num Epochs	10
Training Config Warmup Steps	1000
Training Config Weight Decay	0.01
Vocab Size	64172

Performance Metrics

Evaluation Results

Metric	Value
Baseline Bleu	0.1178
Baseline Chrf	33.0561
Bleu Improvement	0.6561
Bleu Improvement Percent	557.2142
Chrf Improvement	51.5256
Chrf Improvement Percent	155.8731
Epoch	10.0000
Eval Bleu	0.7739
Eval Chrf	84.5818
Eval Loss	0.2106
Eval Runtime	481.4612
Eval Samples Per Second	15.5820
Eval Steps Per Second	3.8960
Final Epoch	10.0000
Final Eval Bleu	0.7739
Final Eval Chrf	84.5818
Final Eval Loss	0.2106
Final Eval Runtime	481.4612
Final Eval Samples Per Second	15.5820
Final Eval Steps Per Second	3.8960
Grad Norm	0.6669
Learning Rate	0.0000
Loss	0.0663
Total Flos	6254416761716736.0000
Total Samples	50012.0000
Train Loss	0.2427
Train Runtime	16184.3599
Train Samples	42510.0000
Train Samples Per Second	26.2660
Train Steps Per Second	6.5670
Validation Samples	7502.0000

Evaluation Results

Metric	Value
Baseline Bleu	0.1178
Baseline Chrf	33.0561
Bleu Improvement	0.6561
Bleu Improvement Percent	557.2142
Chrf Improvement	51.5256
Chrf Improvement Percent	155.8731
Epoch	10.0000
Eval Bleu	0.7739
Eval Chrf	84.5818
Eval Loss	0.2106
Eval Runtime	481.4612
Eval Samples Per Second	15.5820
Eval Steps Per Second	3.8960
Final Epoch	10.0000
Final Eval Bleu	0.7739
Final Eval Chrf	84.5818
Final Eval Loss	0.2106
Final Eval Runtime	481.4612
Final Eval Samples Per Second	15.5820
Final Eval Steps Per Second	3.8960
Grad Norm	0.6669
Learning Rate	0.0000
Loss	0.0663
Total Flos	6254416761716736.0000
Total Samples	50012.0000
Train Loss	0.2427
Train Runtime	16184.3599
Train Samples	42510.0000
Train Samples Per Second	26.2660
Train Steps Per Second	6.5670
Validation Samples	7502.0000

Evaluation Results

Metric	Value
Baseline Bleu	0.1178
Baseline Chrf	33.0561
Bleu Improvement	0.6561
Bleu Improvement Percent	557.2142
Chrf Improvement	51.5256
Chrf Improvement Percent	155.8731
Epoch	10.0000
Eval Bleu	0.7739
Eval Chrf	84.5818
Eval Loss	0.2106
Eval Runtime	481.4612
Eval Samples Per Second	15.5820
Eval Steps Per Second	3.8960
Final Epoch	10.0000
Final Eval Bleu	0.7739
Final Eval Chrf	84.5818
Final Eval Loss	0.2106
Final Eval Runtime	481.4612
Final Eval Samples Per Second	15.5820
Final Eval Steps Per Second	3.8960
Grad Norm	0.6669
Learning Rate	0.0000
Loss	0.0663
Total Flos	6254416761716736.0000
Total Samples	50012.0000
Train Loss	0.2427
Train Runtime	16184.3599
Train Samples	42510.0000
Train Samples Per Second	26.2660
Train Steps Per Second	6.5670
Validation Samples	7502.0000

Evaluation Results

Metric	Value
Baseline Bleu	0.1178
Baseline Chrf	33.0561
Bleu Improvement	0.6561
Bleu Improvement Percent	557.2142
Chrf Improvement	51.5256
Chrf Improvement Percent	155.8731
Epoch	10.0000
Eval Bleu	0.7739
Eval Chrf	84.5818
Eval Loss	0.2106
Eval Runtime	481.4612
Eval Samples Per Second	15.5820
Eval Steps Per Second	3.8960
Final Epoch	10.0000
Final Eval Bleu	0.7739
Final Eval Chrf	84.5818
Final Eval Loss	0.2106
Final Eval Runtime	481.4612
Final Eval Samples Per Second	15.5820
Final Eval Steps Per Second	3.8960
Grad Norm	0.6669
Learning Rate	0.0000
Loss	0.0663
Total Flos	6254416761716736.0000
Total Samples	50012.0000
Train Loss	0.2427
Train Runtime	16184.3599
Train Samples	42510.0000
Train Samples Per Second	26.2660
Train Steps Per Second	6.5670
Validation Samples	7502.0000

Usage

from transformers import pipeline

translator = pipeline("translation", model="marlonbino/opus-mt-lg-en-finetuned")
result = translator("Your text here")
print(result[0]["translation_text"])

MLflow Tracking

• Experiment: translation-lg-en
• Run ID: 9ebbc90beedf44be9c2c886d442a19ef
• Training Date: 2025-07-29 18:20:39
• Tracking URI: http://192.168.10.6:5000

Training Metrics Visualization

View detailed training metrics and TensorBoard logs in the Training metrics tab.

Citation

@misc{opus_mt_lg_en_finetuned,
  title={opus-mt-lg-en-finetuned},
  author={OpenCHS Team},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/marlonbino/opus-mt-lg-en-finetuned}
}

Contact

info@bitz-itc.com

---

Model card auto-generated from MLflow