File size: 1,147 Bytes
34f3d98 b595407 9cb32c2 b595407 |
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 |
---
license: mit
language:
- en
metrics:
- rouge
base_model:
- allenai/led-base-16384
pipeline_tag: summarization
library_name: transformers
---
# Longformer fine-tuned to summarize Terms of Service
Terms of Service documents are lengthy, complex, and time-consuming to read. Due to its vague language, people often don’t understand what they are agreeing to. Hence, we have fine-tuned Longformer model to help summarize TOS and make it easy to read and understand.
<br><br>
This model is a fine-tuned version of [`allenai/led-base-16384`](https://huggingface.co/allenai/led-base-16384) <br>
Dataset used: TL;DRLegal and TOS;DR website <br>
It achieves the following results on the validation set:<br>
* ROUGE-1: 0.28
* ROUGE-2: 0.13
* ROUGE-L: 0.27
## How to Use
```python
from transformers import AutoTokenizer, AutoModelForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained("aarushi-211/TOS-Longformer")
model = AutoModelForSequenceClassification.from_pretrained("aarushi-211/TOS-Longformer")
inputs = tokenizer("Your input text here", return_tensors="pt", truncation=True)
outputs = model(**inputs)
logits = outputs.logits |