bart-large-cnn-with-generate-finetune-indosum
This model is a fine-tuned version of facebook/bart-large-cnn on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0686
- Rouge1: 0.8873
- Rouge2: 0.8491
- Rougel: 0.8815
- Rougelsum: 0.8815
- Gen Len: 128.9129
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: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|---|---|---|---|---|---|---|---|---|
| 0.2591 | 1.0 | 4460 | 0.2573 | 0.7218 | 0.6324 | 0.6969 | 0.6967 | 129.0612 |
| 0.1657 | 2.0 | 8920 | 0.1600 | 0.7613 | 0.6815 | 0.7401 | 0.7401 | 128.9508 |
| 0.0945 | 3.0 | 13380 | 0.1157 | 0.8001 | 0.7311 | 0.7837 | 0.7835 | 128.9105 |
| 0.0508 | 4.0 | 17840 | 0.0976 | 0.8277 | 0.7704 | 0.8152 | 0.8152 | 129.0289 |
| 0.0296 | 5.0 | 22300 | 0.0853 | 0.857 | 0.8087 | 0.8473 | 0.8471 | 128.9257 |
| 0.0176 | 6.0 | 26760 | 0.0793 | 0.8702 | 0.8279 | 0.8632 | 0.8633 | 128.9113 |
| 0.0112 | 7.0 | 31220 | 0.0605 | 0.8789 | 0.8377 | 0.872 | 0.8721 | 128.8637 |
| 0.0074 | 8.0 | 35680 | 0.0597 | 0.88 | 0.84 | 0.8731 | 0.8732 | 128.9305 |
| 0.005 | 9.0 | 40140 | 0.0658 | 0.8822 | 0.8433 | 0.8761 | 0.8761 | 128.949 |
| 0.0036 | 10.0 | 44600 | 0.0686 | 0.8873 | 0.8491 | 0.8815 | 0.8815 | 128.9129 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.0.0
- Datasets 2.12.0
- Tokenizers 0.13.2