| license: apache-2.0 | |
| tags: | |
| - translation | |
| - generated_from_trainer | |
| datasets: | |
| - tatoeba_mt | |
| metrics: | |
| - bleu | |
| model-index: | |
| - name: fine-tuned_ar-en | |
| results: | |
| - task: | |
| name: Sequence-to-sequence Language Modeling | |
| type: text2text-generation | |
| dataset: | |
| name: tatoeba_mt | |
| type: tatoeba_mt | |
| config: ara-eng | |
| split: validation | |
| args: ara-eng | |
| metrics: | |
| - name: Bleu | |
| type: bleu | |
| value: 51.81577100911173 | |
| <!-- This model card has been generated automatically according to the information the Trainer had access to. You | |
| should probably proofread and complete it, then remove this comment. --> | |
| # fine-tuned_ar-en | |
| This model is a fine-tuned version of [Helsinki-NLP/opus-mt-ar-en](https://huggingface.co/Helsinki-NLP/opus-mt-ar-en) on the tatoeba_mt dataset. | |
| It achieves the following results on the evaluation set: | |
| - Loss: 0.8464 | |
| - Bleu: 51.8158 | |
| ## Model description | |
| More information needed | |
| ## 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: 2e-05 | |
| - train_batch_size: 32 | |
| - eval_batch_size: 64 | |
| - seed: 42 | |
| - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 | |
| - lr_scheduler_type: linear | |
| - num_epochs: 3 | |
| - mixed_precision_training: Native AMP | |
| ### Training results | |
| ### Framework versions | |
| - Transformers 4.25.1 | |
| - Pytorch 1.13.0+cu116 | |
| - Datasets 2.8.0 | |
| - Tokenizers 0.13.2 | |