metadata

license: apache-2.0
tags:
  - summarization
  - generated_from_trainer
datasets:
  - xlsum
metrics:
  - rouge
model-index:
  - name: AraBART-finetuned-ar
    results:
      - task:
          name: Sequence-to-sequence Language Modeling
          type: text2text-generation
        dataset:
          name: xlsum
          type: xlsum
          args: arabic
        metrics:
          - name: Rouge1
            type: rouge
            value: 2.2459

AraBART-finetuned-ar

This model is a fine-tuned version of moussaKam/AraBART on the xlsum dataset. It achieves the following results on the evaluation set:

Loss: 2.3785
Rouge1: 2.2459
Rouge2: 0.0
Rougel: 2.2459
Rougelsum: 2.2459
Gen Len: 19.695

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: 5e-05
train_batch_size: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 16
total_train_batch_size: 64
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.6
num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	0.98	32	2.7774	1.0638	0.0	1.0638	1.182	19.5177
No log	1.98	64	2.4730	1.182	0.0	1.3002	1.182	19.8121
No log	2.98	96	2.4129	2.3641	0.3546	2.3641	2.3641	19.8298
No log	3.98	128	2.3724	2.1277	0.3546	2.1277	2.1277	19.8121
No log	4.98	160	2.3560	1.8913	0.3546	1.8913	1.8913	19.805
No log	5.98	192	2.3574	1.5366	0.0	1.5366	1.6548	19.7979
No log	6.98	224	2.3676	2.1277	0.3546	2.2459	2.1277	19.6348
No log	7.98	256	2.3656	2.0095	0.0	2.0095	2.0095	19.844
No log	8.98	288	2.3751	2.2459	0.0	2.3641	2.2459	19.6738
No log	9.98	320	2.3785	2.2459	0.0	2.2459	2.2459	19.695

Framework versions

Transformers 4.17.0
Pytorch 1.10.0+cu111
Datasets 2.0.0
Tokenizers 0.11.6