t5-base-qasper

This model is a fine-tuned version of t5-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.1947
• Answer f1: 0.0483

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	Answer f1
No log	1.0	262	1.4772	0.0433
1.5405	2.0	524	1.2919	0.0492
1.5405	3.0	786	1.2517	0.0491
1.1476	4.0	1048	1.2292	0.0492
1.1476	5.0	1310	1.2197	0.0497
1.056	6.0	1572	1.2150	0.0509
1.056	7.0	1834	1.2116	0.0507
0.9915	8.0	2096	1.2048	0.0503
0.9915	9.0	2358	1.2056	0.0512
0.9418	10.0	2620	1.1954	0.0497
0.9418	11.0	2882	1.1977	0.0491
0.9348	12.0	3144	1.1954	0.0486
0.9348	13.0	3406	1.1926	0.0482
0.9073	14.0	3668	1.1946	0.0486
0.9073	15.0	3930	1.1919	0.0480
0.8769	16.0	4192	1.1955	0.0485
0.8769	17.0	4454	1.1941	0.0481
0.8754	18.0	4716	1.1947	0.0483

Framework versions

• Transformers 4.24.0
• Pytorch 1.13.0+cu117
• Datasets 2.7.1
• Tokenizers 0.13.2