all_2490_bart-base

This model is a fine-tuned version of facebook/bart-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.0206
• Rouge1: 0.2426
• Rouge2: 0.1208
• Rougel: 0.2025
• Rougelsum: 0.2266
• Gen Len: 19.9945

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: 32
• eval_batch_size: 20
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 512
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.7151	0.8	500	1.1257	0.2361	0.1122	0.1955	0.2196	19.9978
1.0837	1.61	1000	1.0810	0.2401	0.1176	0.1997	0.2237	19.9953
1.0348	2.41	1500	1.0651	0.2401	0.1179	0.1999	0.2239	19.9957
1.0059	3.21	2000	1.0522	0.2402	0.1183	0.2001	0.2242	19.996
0.9855	4.02	2500	1.0439	0.2416	0.1197	0.2014	0.2257	19.9948
0.9642	4.82	3000	1.0361	0.2421	0.12	0.2019	0.2263	19.9936
0.9519	5.63	3500	1.0329	0.2415	0.1199	0.2016	0.2258	19.9948
0.9389	6.43	4000	1.0278	0.2424	0.1204	0.2022	0.2265	19.9942
0.9302	7.23	4500	1.0273	0.2422	0.1204	0.2022	0.2264	19.9943
0.9225	8.04	5000	1.0219	0.2421	0.1208	0.2023	0.2263	19.9946
0.9152	8.84	5500	1.0219	0.2429	0.1208	0.2027	0.227	19.9948
0.911	9.64	6000	1.0206	0.2426	0.1208	0.2025	0.2266	19.9945

Framework versions

• Transformers 4.38.2
• Pytorch 2.0.0+cu117
• Datasets 2.18.0
• Tokenizers 0.15.2