metadata
license: apache-2.0
base_model: Helsinki-NLP/opus-mt-de-en
tags:
- generated_from_trainer
metrics:
- bleu
model-index:
- name: spark-name-de-to-en
results: []
German Names to English Translation Model
Model Overview
This translation model is specifically designed to accurately and fluently translate German names and surnames into English.
Intended Uses and Limitations
This model is built for Spark IT enterprise looking to automate the translation process of German names and surnames into English.
Training and Evaluation Data
This model has been trained on a diverse dataset consisting of over 68,493 lines of data, encompassing a wide range of Hindi names and surnames along with their English counterparts. Evaluation data has been carefully selected to ensure reliable and accurate translation performance.
Training Procedure
- 1 days of training
Hardware Environment:
- Azure Studio
- Standard_DS12_v2
- 4 cores, 28GB RAM, 56GB storage
- Data manipulation and training on medium-sized datasets (1-10GB)
- 6 cores
- Loss: 0.4618
- Bleu: 70.7674
- Gen Len: 10.2548
Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
|---|---|---|---|---|---|
| 1.0715 | 1.0 | 1600 | 0.9902 | 41.5519 | 5.8328 |
| 0.8185 | 2.0 | 3200 | 0.9547 | 53.8222 | 5.7988 |
| 0.6909 | 3.0 | 4800 | 0.9527 | 54.7846 | 5.8169 |
| 0.6038 | 4.0 | 6400 | 0.9496 | 55.6009 | 5.8406 |
Framework versions
- Transformers 4.39.1
- Pytorch 2.2.2+cpu
- Datasets 2.15.0
- Tokenizers 0.15.2