<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
bart-large-cnn-samsum-ElectrifAi_v7
This model is a fine-tuned version of philschmid/bart-large-cnn-samsum on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.3396
- Rouge1: 53.6259
- Rouge2: 29.3085
- Rougel: 39.5423
- Rougelsum: 51.4836
- Gen Len: 105.963
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: 4
- eval_batch_size: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
---|---|---|---|---|---|---|---|---|
No log | 1.0 | 24 | 1.3377 | 53.4184 | 29.2938 | 38.7671 | 51.2253 | 114.9259 |
No log | 2.0 | 48 | 1.3225 | 53.1653 | 29.2807 | 40.6062 | 50.9565 | 116.037 |
No log | 3.0 | 72 | 1.3396 | 53.6259 | 29.3085 | 39.5423 | 51.4836 | 105.963 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.12.1
- Datasets 2.6.1
- Tokenizers 0.13.2