Einmalumdiewelt's picture
update model card README.md
9456e52
|
raw
history blame
1.27 kB
metadata
language:
  - de
tags:
  - generated_from_trainer
metrics:
  - rouge
model-index:
  - name: T5-Base_GNAD_MaxSamples
    results: []

T5-Base_GNAD_MaxSamples

This model is a fine-tuned version of Einmalumdiewelt/T5-Base_GNAD_MaxSamples on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 2.0282
  • Rouge1: 28.26
  • Rouge2: 9.2063
  • Rougel: 19.7844
  • Rougelsum: 24.388
  • Gen Len: 53.322

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-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: 10.0

Training results

Framework versions

  • Transformers 4.22.0.dev0
  • Pytorch 1.12.1+cu113
  • Datasets 2.4.0
  • Tokenizers 0.12.1