File size: 2,680 Bytes
bf2f610 5b54958 bf2f610 5b54958 bf2f610 5b54958 bf2f610 5b54958 bf2f610 5b54958 bf2f610 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 |
---
tags:
- generated_from_trainer
metrics:
- rouge
model-index:
- name: bert2gpt2SUMM-finetuned-mlsum-finetuned-mlorange_sum-finetuned_med_sum
results: []
---
<!-- 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. -->
# bert2gpt2SUMM-finetuned-mlsum-finetuned-mlorange_sum-finetuned_med_sum
This model is a fine-tuned version of [Chemsseddine/bert2gpt2SUMM-finetuned-mlsum-finetuned-mlorange_sum](https://huggingface.co/Chemsseddine/bert2gpt2SUMM-finetuned-mlsum-finetuned-mlorange_sum) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 2.0684
- Rouge1: 34.1248
- Rouge2: 17.7006
- Rougel: 33.4661
- Rougelsum: 33.4419
- Gen Len: 22.6429
## 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: 1
- eval_batch_size: 1
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|:-------------:|:-----:|:-----:|:---------------:|:-------:|:-------:|:-------:|:---------:|:-------:|
| 2.9107 | 1.0 | 1000 | 2.0877 | 30.4547 | 14.4024 | 30.3642 | 30.3788 | 21.9714 |
| 1.8782 | 2.0 | 2000 | 1.8151 | 32.6607 | 16.8089 | 32.3844 | 32.4762 | 21.7714 |
| 1.291 | 3.0 | 3000 | 1.7523 | 33.6391 | 16.7866 | 32.4256 | 32.3306 | 22.7429 |
| 0.819 | 4.0 | 4000 | 1.7650 | 35.0633 | 19.1222 | 34.4902 | 34.6796 | 22.4714 |
| 0.4857 | 5.0 | 5000 | 1.8129 | 33.8763 | 16.9303 | 32.8845 | 32.9225 | 22.3857 |
| 0.3232 | 6.0 | 6000 | 1.9339 | 33.9272 | 17.1784 | 32.9301 | 33.0253 | 22.4286 |
| 0.2022 | 7.0 | 7000 | 1.9634 | 33.9869 | 16.4238 | 33.7336 | 33.65 | 22.6429 |
| 0.1452 | 8.0 | 8000 | 2.0090 | 33.8892 | 18.2723 | 33.7514 | 33.6531 | 22.5714 |
| 0.0845 | 9.0 | 9000 | 2.0337 | 33.9649 | 17.1339 | 33.5061 | 33.4157 | 22.7857 |
| 0.0531 | 10.0 | 10000 | 2.0684 | 34.1248 | 17.7006 | 33.4661 | 33.4419 | 22.6429 |
### Framework versions
- Transformers 4.20.1
- Pytorch 1.11.0+cu113
- Datasets 2.3.2
- Tokenizers 0.12.1
|