metadata
tags:
- generated_from_trainer
model-index:
- name: fine-tune-Pegasus-large
results: []
fine-tune-Pegasus-large
This model is a fine-tuned version of google/pegasus-large on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 11.0526
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: 6.35e-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: cosine
- lr_scheduler_warmup_steps: 500
- num_epochs: 1
- mixed_precision_training: Native AMP
Training results
Framework versions
- Transformers 4.15.0
- Pytorch 1.10.1
- Datasets 1.17.0
- Tokenizers 0.10.3