Create README.md

README.md

@@ -0,0 +1,72 @@
+---
+license: cc-by-4.0
+language:
+- en
+pipeline_tag: text-classification
+tags:
+- roberta-large
+- topic
+- news
+
+widget:
+- text: "Diplomatic efforts to deal with the world’s two wars — the civil war in Spain and the undeclared Chinese - Japanese conflict — received sharp setbacks today."
+- text: "WASHINGTON. AP. A decisive development appeared in the offing in the tug-of-war between the federal government and the states over the financing of relief."
+- text: "A frantic bride called the Rochester Gas and Electric corporation to complain that her new refrigerator “freezes ice cubes too fast.”"
+
+---
+
+# Fine-tuned RoBERTa-large for detecting news on obituaries
+
+# Model Description
+
+This model is a finetuned RoBERTa-large, for classifying whether news articles are obituaries. 
+
+# How to Use
+
+```python
+from transformers import pipeline
+classifier = pipeline("text-classification", model="dell-research-harvard/topic-obits")
+classifier("John Smith died after a long illness")
+```
+
+# Training data
+
+The model was trained on a hand-labelled sample of data from the [NEWSWIRE dataset](https://huggingface.co/datasets/dell-research-harvard/newswire).
+
+Split|Size
+-|-
+Train|272
+Dev|57
+Test|57
+
+# Test set results
+
+Metric|Result
+-|-
+F1|1.000
+Accuracy|1.000
+Precision|1.000
+Recall|1.000
+
+
+# Citation Information
+
+You can cite this dataset using
+
+```
+@misc{silcock2024newswirelargescalestructureddatabase,
+      title={Newswire: A Large-Scale Structured Database of a Century of Historical News}, 
+      author={Emily Silcock and Abhishek Arora and Luca D'Amico-Wong and Melissa Dell},
+      year={2024},
+      eprint={2406.09490},
+      archivePrefix={arXiv},
+      primaryClass={cs.CL},
+      url={https://arxiv.org/abs/2406.09490}, 
+}
+```
+
+# Applications
+
+We applied this model to a century of historical news articles. You can see all the classifications in the [NEWSWIRE dataset](https://huggingface.co/datasets/dell-research-harvard/newswire). 
+
+