PTS-Bart-Large-CNN / README.md
ahmedmbutt's picture
Update README.md
981f027 verified
|
raw
history blame
3.39 kB
---
license: mit
base_model: facebook/bart-large-cnn
tags:
- generated_from_trainer
metrics:
- rouge
model-index:
- name: PTS-Bart-Large-CNN
results:
- task:
type: summarization
name: Summarization
dataset:
name: PTS Dataset
type: PTS-Dataset
metrics:
- name: Rouge1
type: rouge
value: 0.6376
- name: Rouge2
type: rouge
value: 0.4143
- name: Rougel
type: rouge
value: 0.538
- name: Rougelsum
type: rouge
value: 0.5387
pipeline_tag: summarization
datasets:
- ahmedmbutt/PTS-Dataset
language:
- en
library_name: transformers
widget:
- text: "I have to say that I do miss talking to a good psychiatrist- however. I could sit and argue for ages with a psychiatrist who is intelligent and kind (quite hard to find- but they do exist). Especially now that I have a PhD in philosophy and have read everything that can be found on madness- including the notes they wrote about me when I was in the hospital. Nowadays- psychiatrists have a tendency to sign me off pretty quickly when I come onto their radar. They don’t wish to deal with me- I tire them out."
---
<!-- 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. -->
# PTS-Bart-Large-CNN
This model is a fine-tuned version of [facebook/bart-large-cnn](https://huggingface.co/facebook/bart-large-cnn) on the PTS dataset.
It achieves the following results on the evaluation set:
- Loss: 1.2638
- Rouge1: 0.6376
- Rouge2: 0.4143
- Rougel: 0.538
- Rougelsum: 0.5387
- Gen Len: 76.8417
<!-- ## 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
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 8
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|:------:|:---------:|:-------:|
| No log | 1.0 | 180 | 0.8748 | 0.6166 | 0.3827 | 0.5058 | 0.5055 | 77.6583 |
| No log | 2.0 | 360 | 0.8774 | 0.6307 | 0.4064 | 0.5302 | 0.531 | 77.5111 |
| 0.6761 | 3.0 | 540 | 0.9064 | 0.635 | 0.4052 | 0.5309 | 0.5311 | 76.2833 |
| 0.6761 | 4.0 | 720 | 1.0386 | 0.6329 | 0.4038 | 0.5261 | 0.5262 | 78.4889 |
| 0.6761 | 5.0 | 900 | 1.0993 | 0.6285 | 0.4016 | 0.5239 | 0.5246 | 77.0083 |
| 0.2016 | 6.0 | 1080 | 1.2025 | 0.6351 | 0.4126 | 0.5351 | 0.5356 | 76.0722 |
| 0.2016 | 7.0 | 1260 | 1.2399 | 0.6356 | 0.4108 | 0.5362 | 0.5368 | 78.5361 |
| 0.2016 | 8.0 | 1440 | 1.2638 | 0.6376 | 0.4143 | 0.538 | 0.5387 | 76.8417 |
### Framework versions
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1