metadata
license: apache-2.0
tags:
- summarization
- mt5
- ar
- abstractive summarization
- generated_from_trainer
datasets:
- wiki_lingua
model-index:
- name: t5-arabic-base-finetuned-wikilingua-ar
results: []
t5-arabic-base-finetuned-wikilingua-ar
This model is a fine-tuned version of bakrianoo/t5-arabic-base on the wiki_lingua dataset. It achieves the following results on the evaluation set:
- Loss: 3.2735
- Rouge-1: 20.72
- Rouge-2: 7.63
- Rouge-l: 18.75
- Gen Len: 18.74
- Bertscore: 70.79
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: 0.0005
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 250
- num_epochs: 8
- label_smoothing_factor: 0.1
Training results
Framework versions
- Transformers 4.19.3
- Pytorch 1.11.0+cu113
- Datasets 2.2.2
- Tokenizers 0.12.1