<!-- 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_v8
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: 1.5051
- Rouge1: 51.3789
- Rouge2: 27.7926
- Rougel: 38.0256
- Rougelsum: 50.0685
- Gen Len: 104.4
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 | 27 | 1.4978 | 54.3415 | 30.2195 | 39.7514 | 52.7012 | 103.2667 |
| No log | 2.0 | 54 | 1.5002 | 51.6735 | 28.1006 | 37.5005 | 49.9317 | 117.8 |
| No log | 3.0 | 81 | 1.5051 | 51.3789 | 27.7926 | 38.0256 | 50.0685 | 104.4 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.12.1
- Datasets 2.6.1
- Tokenizers 0.13.2
