README.md

---
license: apache-2.0
datasets:
- ClimatePolicyRadar/national-climate-targets
language:
- en
pipeline_tag: text-classification
tags:
- climate
widget:
- text: "The Net Zero Strategy, published in October 2021, was the first document of its kind for a major economy. It set out the government’s vision for a market-led, technology-driven transition to decarbonise the UK economy and reach net zero by 2050."
inference:
  parameters:
    function_to_apply: "sigmoid"
---

## National Climate Targets Classifier - Climate Policy Radar

A multi-label text-classifier trained on the National Climate Targets dataset by Climate Policy Radar.

Using the [climatebert/distilroberta-base-climate-f](https://huggingface.co/climatebert/distilroberta-base-climate-f) model as a starting point, this classifier is trained on the [ClimatePolicyRadar/national-climate-targets](https://huggingface.co/datasets/ClimatePolicyRadar/national-climate-targets) dataset to predict Net Zero ("NZT")
, "Reduction" and "Other" targets in a multi-label setting. The training data is an expert annotated subset of national laws, policies and UNFCCC submissions.


For more information on the annotation methodology and classifier training [see our paper](https://arxiv.org/abs/2404.02822).

## Getting started


```python
from transformers import AutoModelForSequenceClassification, AutoTokenizer


model_name = "ClimatePolicyRadar/national-climate-targets"
example = "The Net Zero Strategy, published in October 2021, was the first "\
    "document of its kind for a major economy. It set out the government’s "\
    "vision for a market-led, technology-driven transition to decarbonise "\
    "the UK economy and reach net zero by 2050."


model = AutoModelForSequenceClassification.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

# using sigmoid because the model is multi-label
pipe = pipeline("text-classification", model=model, tokenizer=tokenizer, function_to_apply="sigmoid")


pipe(example, padding=True, truncation=True, return_all_scores=True)

>>> [[{'label': 'NZT', 'score': 0.9142044186592102},
  {'label': 'Reduction', 'score': 0.04552844911813736},
  {'label': 'Other', 'score': 0.07590094953775406}]]
```


## Licence

Our classifier is licensed as [Apache 2.0](https://www.apache.org/licenses/LICENSE-2.0).

Please read our [Terms of Use](https://app.climatepolicyradar.org/terms-of-use), including any specific terms relevant to commercial use. Contact [email protected] with any questions.


## Links

- [Paper](https://arxiv.org/abs/2404.02822)


## Citation

```
@misc{juhasz2024identifying,
      title={Identifying Climate Targets in National Laws and Policies using Machine Learning}, 
      author={Matyas Juhasz and Tina Marchand and Roshan Melwani and Kalyan Dutia and Sarah Goodenough and Harrison Pim and Henry Franks},
      year={2024},
      eprint={2404.02822},
      archivePrefix={arXiv},
      primaryClass={cs.CY}
}
```

## Authors & Contact
Climate Policy Radar team: Matyas Juhasz, Tina Marchand, Roshan Melwani, Kalyan Dutia, Sarah Goodenough, Harrison Pim, and Henry Franks.

[email protected]

https://climatepolicyradar.org