Update README.md
Browse files
README.md
CHANGED
|
@@ -18,6 +18,9 @@ datasets: "fhamborg/news_sentiment_newsmtsc"
|
|
| 18 |
|
| 19 |
# NewsSentiment: easy-to-use, high-quality target-dependent sentiment classification for news articles
|
| 20 |
|
|
|
|
|
|
|
|
|
|
| 21 |
This model is the currently [best performing](https://aclanthology.org/2021.eacl-main.142.pdf)
|
| 22 |
targeted sentiment classifier for news articles. In contrast to regular sentiment
|
| 23 |
classification, targeted sentiment classification allows you to provide a target in a sentence.
|
|
|
|
| 18 |
|
| 19 |
# NewsSentiment: easy-to-use, high-quality target-dependent sentiment classification for news articles
|
| 20 |
|
| 21 |
+
## Important: use our PyPI package!
|
| 22 |
+
The Huggingface Hub architecture currently [does not support](https://github.com/huggingface/transformers/issues/14785) target-dependent sentiment classification since one cannot provide the required inputs, i.e., sentence and target. Thus, we recommend that you use our easy-to-use [PyPI package NewsSentiment](https://pypi.org/project/NewsSentiment/).
|
| 23 |
+
|
| 24 |
This model is the currently [best performing](https://aclanthology.org/2021.eacl-main.142.pdf)
|
| 25 |
targeted sentiment classifier for news articles. In contrast to regular sentiment
|
| 26 |
classification, targeted sentiment classification allows you to provide a target in a sentence.
|