res_nw_dj

This model is a fine-tuned version of riotu-lab/ArabianGPT-01B on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6246
• Bleu: 0.3877
• Rouge1: 0.5958
• Rouge2: 0.3370
• Rougel: 0.5935

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 20.0

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Rouge1	Rouge2	Rougel
1.2336	1.0	2679	0.7062	0.3526	0.5198	0.2547	0.5170
0.634	2.0	5358	0.6423	0.3756	0.5739	0.3114	0.5714
0.5299	3.0	8037	0.6246	0.3877	0.5958	0.3370	0.5935
0.4492	4.0	10716	0.6246	0.3905	0.6081	0.3526	0.6057
0.3829	5.0	13395	0.6300	0.3963	0.6145	0.3621	0.6125
0.328	6.0	16074	0.6384	0.3961	0.6213	0.3700	0.6194
0.2832	7.0	18753	0.6491	0.3999	0.6232	0.3741	0.6209
0.2453	8.0	21432	0.6607	0.3968	0.6232	0.3746	0.6212

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.3.1+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1