bart-abs-2409-1947-lr-3e-05-bs-2-maxep-10

This model is a fine-tuned version of sshleifer/distilbart-xsum-12-6 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.6542
• Rouge/rouge1: 0.4733
• Rouge/rouge2: 0.2228
• Rouge/rougel: 0.409
• Rouge/rougelsum: 0.4103
• Bertscore/bertscore-precision: 0.8957
• Bertscore/bertscore-recall: 0.8945
• Bertscore/bertscore-f1: 0.8949
• Meteor: 0.4257
• Gen Len: 38.0727

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: 3e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge/rouge1	Rouge/rouge2	Rouge/rougel	Rouge/rougelsum	Bertscore/bertscore-precision	Bertscore/bertscore-recall	Bertscore/bertscore-f1	Meteor	Gen Len
2.3299	1.0	434	2.1997	0.4484	0.2127	0.3842	0.3856	0.8939	0.8904	0.892	0.4107	38.4273
1.6174	2.0	868	2.0581	0.4617	0.2158	0.3919	0.393	0.8972	0.8907	0.8938	0.4049	36.0545
1.1854	3.0	1302	2.1888	0.4631	0.2153	0.3893	0.3914	0.9	0.891	0.8953	0.4044	35.8273
0.8568	4.0	1736	2.3776	0.4518	0.201	0.3822	0.3839	0.8973	0.8895	0.8932	0.3907	34.7636
0.6056	5.0	2170	2.6468	0.4731	0.2207	0.4031	0.4036	0.8974	0.8946	0.8959	0.4173	37.5818
0.4235	6.0	2604	2.9226	0.4816	0.2258	0.4055	0.4064	0.8989	0.8955	0.897	0.4317	36.8909
0.2995	7.0	3038	3.1938	0.4541	0.1989	0.3839	0.3843	0.8941	0.8916	0.8927	0.4071	37.0091
0.2079	8.0	3472	3.4178	0.4684	0.2094	0.3926	0.3931	0.8941	0.8942	0.894	0.4177	39.4545
0.1527	9.0	3906	3.5347	0.4716	0.2151	0.3989	0.4009	0.8937	0.8942	0.8938	0.4277	39.5909
0.1203	10.0	4340	3.6542	0.4733	0.2228	0.409	0.4103	0.8957	0.8945	0.8949	0.4257	38.0727

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0
• Datasets 3.0.0
• Tokenizers 0.19.1