metadata
base_model: UBC-NLP/AraT5v2-base-1024
tags:
- generated_from_trainer
datasets:
- opus100
metrics:
- bleu
model-index:
- name: finetune-t5-base-on-opus100-Ar2En-without-optimization
results:
- task:
name: Sequence-to-sequence Language Modeling
type: text2text-generation
dataset:
name: opus100
type: opus100
config: ar-en
split: train[:7000]
args: ar-en
metrics:
- name: Bleu
type: bleu
value: 10.4288
finetune-t5-base-on-opus100-Ar2En-without-optimization
This model is a fine-tuned version of UBC-NLP/AraT5v2-base-1024 on the opus100 dataset. It achieves the following results on the evaluation set:
- Loss: 3.0042
- Bleu: 10.4288
- Gen Len: 10.739
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: 10
- eval_batch_size: 10
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 18
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
|---|---|---|---|---|---|
| 10.1448 | 1.0 | 210 | 3.9256 | 2.8335 | 9.4988 |
| 4.9822 | 2.0 | 420 | 3.5760 | 4.9001 | 10.3329 |
| 4.42 | 3.0 | 630 | 3.4037 | 5.6973 | 10.301 |
| 4.1414 | 4.0 | 840 | 3.3057 | 6.5224 | 10.5559 |
| 3.9451 | 5.0 | 1050 | 3.2169 | 7.409 | 10.7571 |
| 3.7972 | 6.0 | 1260 | 3.1759 | 8.1445 | 10.5908 |
| 3.6687 | 7.0 | 1470 | 3.1340 | 8.246 | 10.7451 |
| 3.5494 | 8.0 | 1680 | 3.1098 | 8.5656 | 10.7616 |
| 3.4748 | 9.0 | 1890 | 3.0749 | 9.052 | 10.8798 |
| 3.3945 | 10.0 | 2100 | 3.0725 | 9.3223 | 10.6794 |
| 3.314 | 11.0 | 2310 | 3.0511 | 9.67 | 10.6871 |
| 3.2606 | 12.0 | 2520 | 3.0398 | 9.6105 | 10.6531 |
| 3.2314 | 13.0 | 2730 | 3.0211 | 10.0661 | 10.752 |
| 3.1557 | 14.0 | 2940 | 3.0188 | 10.0724 | 10.7188 |
| 3.1571 | 15.0 | 3150 | 3.0148 | 10.3648 | 10.7596 |
| 3.1213 | 16.0 | 3360 | 3.0061 | 10.4008 | 10.7784 |
| 3.1111 | 17.0 | 3570 | 3.0077 | 10.4588 | 10.7155 |
| 3.0851 | 18.0 | 3780 | 3.0042 | 10.4288 | 10.739 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.0.0
- Datasets 2.1.0
- Tokenizers 0.15.0