bart-base-aeslc-10-cnt-supervised-basic

This model is a fine-tuned version of facebook/bart-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 5.2943
• Rouge1: 0.1696
• Rouge2: 0.0823
• Rougel: 0.1631
• Rougelsum: 0.1636

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: 1e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.06
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
5.7302	1.6667	5	7.1464	0.158	0.0706	0.1462	0.1466
4.8072	3.3333	10	6.0947	0.1584	0.0706	0.1465	0.1471
3.458	5.0	15	5.3886	0.1584	0.0722	0.147	0.1476
2.8224	6.6667	20	5.1206	0.1532	0.071	0.144	0.1442
2.6376	8.3333	25	5.0738	0.1528	0.0755	0.1456	0.1455
2.3504	10.0	30	5.1150	0.1486	0.0748	0.1437	0.1438
1.7316	11.6667	35	5.1763	0.1515	0.0754	0.1465	0.1465
1.7152	13.3333	40	5.2225	0.1563	0.0768	0.1509	0.1511
1.5365	15.0	45	5.2511	0.1657	0.0804	0.1597	0.1601
1.5521	16.6667	50	5.2756	0.1682	0.0807	0.1618	0.1621
1.2753	18.3333	55	5.2903	0.1691	0.0812	0.1625	0.163
1.4062	20.0	60	5.2943	0.1696	0.0823	0.1631	0.1636

Framework versions

• Transformers 4.57.3
• Pytorch 2.9.1+cu128
• Datasets 3.6.0
• Tokenizers 0.22.1