bart-base-task2

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: 3.6252
• Rouge L: 32.9410
• Bleu-4: 14.4073
• Bertscore F1: 87.7711

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge L	Bleu-4	Bertscore F1
0.1214	1.0	246	3.5625	36.9964	17.5260	88.3887
0.1675	2.0	492	3.5769	36.3781	17.7675	88.1171
0.1537	3.0	738	3.5597	36.5667	18.8463	88.1896
0.2611	4.0	984	3.5809	36.8491	18.0227	88.3404
0.2757	5.0	1230	3.5272	37.0400	18.8248	88.3174
0.1753	6.0	1476	3.5108	37.3436	18.9362	88.3913
0.1228	7.0	1722	3.5914	35.9171	18.1235	88.1545
0.1068	8.0	1968	3.6325	36.8064	17.8291	88.2906
0.0967	9.0	2214	3.6382	37.9565	18.6469	88.3295
0.0857	10.0	2460	3.6406	37.1544	18.5282	88.2495

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.3.1
• Tokenizers 0.21.0