bart-pt-asqa-ob

This model is a fine-tuned version of vblagoje/bart_lfqa on the ASQA dataset. It achieves the following results on the evaluation set:

• Loss: 1.6901
• Rougelsum: 20.7527

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-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rougelsum
No log	1.0	355	1.6295	17.7502
1.6407	2.0	710	1.6144	18.5897
1.4645	3.0	1065	1.6222	19.3778
1.4645	4.0	1420	1.6522	19.6941
1.3678	5.0	1775	1.6528	20.3110
1.2671	6.0	2130	1.6879	20.6112
1.2671	7.0	2485	1.6901	20.7527

Framework versions

• Transformers 4.23.0.dev0
• Pytorch 1.12.1+cu102
• Datasets 2.4.0
• Tokenizers 0.12.1