File size: 1,954 Bytes
1034a01 |
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 |
---
license: mit
base_model: facebook/bart-large-xsum
tags:
- generated_from_trainer
metrics:
- rouge
model-index:
- name: bart_samsum
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. -->
# bart_samsum
This model is a fine-tuned version of [facebook/bart-large-xsum](https://huggingface.co/facebook/bart-large-xsum) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.5520
- Rouge1: 53.0152
- Rouge2: 28.029
- Rougel: 43.8864
- Rougelsum: 48.7945
- Gen Len: 30.3138
## 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: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- 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 |
|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|:-------:|:---------:|:-------:|
| 1.4812 | 1.0 | 921 | 1.5084 | 52.6238 | 27.9144 | 43.3378 | 48.77 | 30.4225 |
| 1.1068 | 2.0 | 1842 | 1.4507 | 53.2142 | 28.5004 | 44.382 | 49.228 | 28.6264 |
| 0.9224 | 3.0 | 2763 | 1.5031 | 52.8334 | 27.9492 | 43.7939 | 48.714 | 29.5751 |
| 0.7957 | 4.0 | 3684 | 1.5520 | 53.0152 | 28.029 | 43.8864 | 48.7945 | 30.3138 |
### Framework versions
- Transformers 4.44.0
- Pytorch 2.4.0
- Datasets 2.21.0
- Tokenizers 0.19.1
|