Switch Transformer (base-8) fine-tuned on samsum dataset for conversation summarization
This model is a fine-tuned version of google/switch-base-8 on the samsum dataset. It achieves the following results on the evaluation set:
- Loss: 1.4614
- Rouge1: 46.1297
- Rouge2: 22.9128
- Rougel: 39.153
- Rougelsum: 42.8502
- Gen Len: 16.9719
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: 5e-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: 6
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|---|---|---|---|---|---|---|---|---|
| 1.8874 | 1.0 | 3683 | 1.5210 | 45.7651 | 22.9379 | 38.8554 | 42.6269 | 17.2482 |
| 1.6301 | 2.0 | 7366 | 1.4628 | 47.2719 | 24.8976 | 40.3913 | 43.9285 | 16.8362 |
| 1.4326 | 3.0 | 11049 | 1.4402 | 47.8275 | 25.2262 | 40.617 | 44.2948 | 16.9523 |
| 1.2992 | 4.0 | 14732 | 1.4489 | 48.393 | 25.3888 | 40.9534 | 44.797 | 17.1504 |
| 1.2259 | 5.0 | 18415 | 1.4495 | 49.2186 | 26.312 | 41.721 | 45.5087 | 17.1956 |
| 1.1477 | 6.0 | 22098 | 1.4610 | 49.0018 | 26.3474 | 41.5217 | 45.4081 | 17.0782 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu116
- Datasets 2.8.0
- Tokenizers 0.13.2