barthez-orange-ft

This model is a fine-tuned version of moussaKam/barthez-orangesum-abstract on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1689
• Rouge1: 0.6719
• Rouge2: 0.6536
• Rougel: 0.6719
• Rougelsum: 0.6722
• Gen Len: 20.0

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: 16
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	31	4.6662	0.6719	0.6535	0.6718	0.6721	20.0
No log	1.99	62	0.6939	0.6718	0.6535	0.6718	0.6721	20.0
No log	2.99	93	0.2939	0.6718	0.6535	0.6718	0.6721	20.0
No log	3.98	124	0.2089	0.6719	0.6535	0.6718	0.6721	20.0
No log	4.98	155	0.1880	0.6719	0.6535	0.6718	0.6721	20.0
No log	5.98	186	0.1795	0.6719	0.6535	0.6718	0.6721	20.0
No log	6.97	217	0.1752	0.6719	0.6535	0.6718	0.6721	20.0
No log	8.0	249	0.1732	0.6719	0.6535	0.6718	0.6721	20.0
No log	9.0	280	0.1716	0.6719	0.6536	0.6719	0.6722	20.0
No log	9.99	311	0.1707	0.6719	0.6536	0.6719	0.6722	20.0
No log	10.99	342	0.1704	0.6719	0.6536	0.6719	0.6722	20.0
No log	11.98	373	0.1696	0.6719	0.6536	0.6719	0.6722	20.0
No log	12.98	404	0.1698	0.6719	0.6536	0.6719	0.6722	20.0
No log	13.98	435	0.1695	0.6719	0.6536	0.6719	0.6722	20.0
No log	14.97	466	0.1693	0.6719	0.6536	0.6719	0.6722	20.0
No log	16.0	498	0.1691	0.6719	0.6536	0.6719	0.6722	20.0
0.9743	17.0	529	0.1691	0.6719	0.6536	0.6719	0.6722	20.0
0.9743	17.99	560	0.1690	0.6719	0.6536	0.6719	0.6722	20.0
0.9743	18.99	591	0.1689	0.6719	0.6536	0.6719	0.6722	20.0
0.9743	19.92	620	0.1689	0.6719	0.6536	0.6719	0.6722	20.0

Framework versions

• Transformers 4.31.0
• Pytorch 2.1.2
• Datasets 2.18.0
• Tokenizers 0.13.3