metadata
license: mit
tags:
- generated_from_trainer
datasets:
- billsum
metrics:
- rouge
model-index:
- name: bart-large-cnn-small-billsum-5epochs
results:
- task:
name: Sequence-to-sequence Language Modeling
type: text2text-generation
dataset:
name: billsum
type: billsum
config: default
split: train[:1%]
args: default
metrics:
- name: Rouge1
type: rouge
value: 0.5406
bart-large-cnn-small-billsum-5epochs
This model is a fine-tuned version of facebook/bart-large-cnn on the billsum dataset. It achieves the following results on the evaluation set:
- Loss: 1.7206
- Rouge1: 0.5406
- Rouge2: 0.312
- Rougel: 0.3945
- Rougelsum: 0.4566
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: 3.373e-05
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- 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
- lr_scheduler_warmup_steps: 16
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum |
|---|---|---|---|---|---|---|---|
| 2.3723 | 1.33 | 16 | 1.8534 | 0.5204 | 0.299 | 0.3893 | 0.4441 |
| 1.6579 | 2.67 | 32 | 1.7208 | 0.5427 | 0.3143 | 0.3915 | 0.459 |
| 1.2397 | 4.0 | 48 | 1.7206 | 0.5406 | 0.312 | 0.3945 | 0.4566 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2