zbigi's picture
End of training
e90b7f9 verified
|
raw
history blame
2.22 kB
metadata
base_model: facebook/bart-large
library_name: peft
license: apache-2.0
metrics:
  - rouge
tags:
  - generated_from_trainer
model-index:
  - name: bart-large-summarization-medical_on_cnn-43
    results: []

bart-large-summarization-medical_on_cnn-43

This model is a fine-tuned version of facebook/bart-large on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 3.0298
  • Rouge1: 0.2443
  • Rouge2: 0.0871
  • Rougel: 0.193
  • Rougelsum: 0.2171
  • Gen Len: 18.859

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: 3e-05
  • train_batch_size: 4
  • eval_batch_size: 1
  • seed: 43
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 6
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Rouge1 Rouge2 Rougel Rougelsum Gen Len
2.2146 1.0 1250 3.0347 0.2365 0.083 0.1868 0.2101 19.329
2.1322 2.0 2500 3.0354 0.2419 0.0862 0.1911 0.2142 19.0
2.0892 3.0 3750 3.0422 0.2411 0.0851 0.1903 0.2134 18.943
2.0772 4.0 5000 3.0387 0.2423 0.0857 0.1911 0.2145 18.869
2.0742 5.0 6250 3.0307 0.2448 0.0868 0.193 0.2171 18.828
2.0673 6.0 7500 3.0298 0.2443 0.0871 0.193 0.2171 18.859

Framework versions

  • PEFT 0.11.1
  • Transformers 4.42.4
  • Pytorch 2.3.1+cu121
  • Datasets 2.20.0
  • Tokenizers 0.19.1