|
--- |
|
license: mit |
|
tags: |
|
- generated_from_trainer |
|
- summarization |
|
datasets: |
|
- orange_sum |
|
model-index: |
|
- name: BART-CNN-Orangesum |
|
results: [] |
|
language: |
|
- fr |
|
- en |
|
--- |
|
|
|
# BART-CNN-Orangesum |
|
|
|
This model is a fine-tuned version of [facebook/bart-large-cnn](https://huggingface.co/facebook/bart-large-cnn) on the orange_sum dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 1.6370 |
|
|
|
It aims at improving the quality of the summary generated on French texts |
|
|
|
## Model description |
|
|
|
this is a fine tuning of the model 'facebook/bart-large-cnn' on the 'orange_sum' dataset |
|
gives better results in French while keeping the intrinsic qualities of the BART model |
|
|
|
## 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: 1 |
|
- eval_batch_size: 1 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 16 |
|
- total_train_batch_size: 16 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_steps: 500 |
|
- num_epochs: 1 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 1.9062 | 0.37 | 500 | 1.8412 | |
|
| 1.6596 | 0.75 | 1000 | 1.6370 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.28.1 |
|
- Pytorch 2.0.0+cu117 |
|
- Datasets 2.12.0 |
|
- Tokenizers 0.13.3 |
