t5-base-asqa-ob / README.md
irenepap's picture
Update README.md
ccf748e
|
raw
history blame
2.26 kB
metadata
license: apache-2.0
tags:
  - generated_from_trainer
model-index:
  - name: t5-base-asqa-ob
    results: []

t5-base-asqa-ob

This model is a fine-tuned version of t5-base on the ASQA dataset with context (open book). It achieves the following results on the evaluation set:

  • Loss: 1.7256
  • Rougelsum: 13.6463

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 50

Training results

Training Loss Epoch Step Validation Loss Rougelsum
2.5404 1.0 710 1.8160 13.0967
2.0048 2.0 1420 1.7752 13.2823
1.9116 3.0 2130 1.7574 13.3068
1.8722 4.0 2840 1.7469 13.3896
1.8298 5.0 3550 1.7395 13.4231
1.8397 6.0 4260 1.7347 13.5553
1.7575 7.0 4970 1.7303 13.5613
1.7433 8.0 5680 1.7266 13.5253
1.7502 9.0 6390 1.7254 13.5391
1.731 10.0 7100 1.7233 13.4958
1.6788 11.0 7810 1.7250 13.5977
1.6793 12.0 8520 1.7243 13.5956
1.6531 13.0 9230 1.7255 13.6186
1.683 14.0 9940 1.7259 13.6567
1.6348 15.0 10650 1.7256 13.6463

Framework versions

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