t5-base-asqa-ob

This model is a fine-tuned version of t5-base on the ASQA dataset. It achieves the following results on the evaluation set:

• Loss: 1.7356
• Rougelsum: 12.0879

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: 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	1.8545	11.6549
2.4887	2.0	710	1.8050	11.7533
1.9581	3.0	1065	1.7843	11.8327
1.9581	4.0	1420	1.7722	11.9442
1.9252	5.0	1775	1.7648	11.9331
1.8853	6.0	2130	1.7567	11.9788
1.8853	7.0	2485	1.7519	12.0300
1.8512	8.0	2840	1.7483	12.0225
1.8328	9.0	3195	1.7451	12.0402
1.8115	10.0	3550	1.7436	12.0444
1.8115	11.0	3905	1.7419	12.0850
1.7878	12.0	4260	1.7408	12.1047
1.774	13.0	4615	1.7394	12.0839
1.774	14.0	4970	1.7390	12.0910
1.7787	15.0	5325	1.7381	12.0880
1.7632	16.0	5680	1.7380	12.1088
1.7623	17.0	6035	1.7370	12.1046
1.7623	18.0	6390	1.7368	12.0997
1.7508	19.0	6745	1.7359	12.0902
1.7597	20.0	7100	1.7356	12.0879

Framework versions

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