README.md

---
language:
- en
license: apache-2.0
datasets:
- xsum
metrics:
- rouge
model-index:
- name: t5-base-finetuned-xsum
  results:
  - task:
      name: Text Summarization
      type: text-summarization
    dataset:
      name: Xsum
      type: xsum
      args: xsum
    metrics:
    - name: rouge
      type: rouge
      value: 0.3414
---


# gpt2-finetuned-xsum

<!-- Provide a quick summary of what the model is/does. -->

This model is t5-base fine-tuned on Xsum dataset for text summarization. 


## Model Details
T5 is an encoder-decoder model pre-trained on a multi-task mixture of unsupervised and supervised tasks and for which each task is converted into a text-to-text format.

## Training Procedure
To train the T5 model for text-summarization, I have used "summarize" prefix before every sentence and gave the encoding of this sentence as input ids and attention mask.
For the labels, I used the encoding of the summaries as the decoder input ids and decoder attention mask.



## Usage:
For generating summaries on a example use: 
```python
predictions = []
tokenised_dataset = tokenizer(documents, truncation=True, padding='max_length', max_length=1024, return_tensors='pt')
source_ids = tokenised_dataset['input_ids']
source_mask = tokenised_dataset['attention_mask']
output = model.generate(input_ids=source_ids, attention_mask=source_mask, max_length=256)
print(tokenizer.decode(output[0], skip_special_tokens=True))
```

## Experiments
We report the ROUGE-1, ROUGE-2 and ROUGE-L on the test datasets.

### Xsum
| ROUGE-1 | ROUGE-2| ROUGE-L|
|---------|--------|--------|
|  0.3414 | 0.1260 | 0.2832 |