bert_large_cnn_daily
This model is a fine-tuned version of alexdg19/bert_large_xsum_samsum2 on the cnn_dailymail dataset. It achieves the following results on the evaluation set:
- Loss: 1.7065
- Rouge1: 0.4251
- Rouge2: 0.2024
- Rougel: 0.2992
- Rougelsum: 0.3961
- Gen Len: 60.6232
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: 3
- eval_batch_size: 3
- seed: 42
- gradient_accumulation_steps: 3
- total_train_batch_size: 9
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 4
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|---|---|---|---|---|---|---|---|---|
| 1.6632 | 1.0 | 1021 | 1.6262 | 0.4191 | 0.1992 | 0.2957 | 0.39 | 60.6205 |
| 1.3734 | 2.0 | 2042 | 1.6078 | 0.4253 | 0.2046 | 0.3009 | 0.397 | 61.0692 |
| 1.1497 | 3.0 | 3064 | 1.6759 | 0.4254 | 0.2033 | 0.2998 | 0.3967 | 60.8555 |
| 1.0123 | 4.0 | 4084 | 1.7065 | 0.4251 | 0.2024 | 0.2992 | 0.3961 | 60.6232 |
Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1