led-risalah_data_v15

This model is a fine-tuned version of silmi224/finetune-led-35000 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.6673
• Rouge1 Precision: 0.7043
• Rouge1 Recall: 0.1227
• Rouge1 Fmeasure: 0.2067

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: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 300
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1 Precision	Rouge1 Recall	Rouge1 Fmeasure
3.0403	1.0	20	2.4986	0.5024	0.0555	0.0987
2.5716	2.0	40	2.1700	0.5606	0.0817	0.1409
2.2879	3.0	60	2.0072	0.5705	0.0869	0.1492
2.0807	4.0	80	1.9094	0.6048	0.0899	0.1542
1.927	5.0	100	1.8184	0.5472	0.0922	0.1561
1.8368	6.0	120	1.7721	0.6079	0.1036	0.1751
1.7468	7.0	140	1.7310	0.639	0.1095	0.1842
1.5913	8.0	160	1.6907	0.6637	0.1109	0.1875
1.534	9.0	180	1.6843	0.6355	0.1102	0.1851
1.4835	10.0	200	1.6605	0.6596	0.1141	0.1922
1.4958	11.0	220	1.6403	0.6929	0.1162	0.1973
1.4547	12.0	240	1.6347	0.6781	0.1118	0.1892
1.3069	13.0	260	1.6604	0.6626	0.1101	0.187
1.2639	14.0	280	1.6712	0.697	0.1227	0.2061
1.3249	15.0	300	1.6255	0.6529	0.1135	0.1914
1.185	16.0	320	1.6484	0.6806	0.1174	0.1981
1.1087	17.0	340	1.6425	0.682	0.1195	0.2008
1.1125	18.0	360	1.6509	0.7122	0.1235	0.2086
1.1574	19.0	380	1.6740	0.6983	0.1214	0.2052
0.9968	20.0	400	1.6673	0.7043	0.1227	0.2067

Framework versions

• Transformers 4.41.2
• Pytorch 2.1.2
• Datasets 2.19.2
• Tokenizers 0.19.1