3 ms·
I would say it's garbage if there's a paper like this: https://arxiv.org/abs/1902.01007 https://arxiv.org/abs/1902.01007
by dunefox 5y ago
I would say it's garbage if there's a paper like this: https://arxiv.org/abs/1902.01007 https://arxiv.org/abs/1902.01007
- visarga 5y agoThat's like hard negative mining. You can also train a weak model and filter out the examples it manages to predict correctly to come up with a harder dataset.
- zibzab 5y agoThat looks interesting. Any more papers on the subject that you can recommend?
- dunefox 5y agoAt the moment, no, but this could be helpful: https://www.connectedpapers.com/main/42ed4a9994e6121a9f325f5b901c5b3d7ce104f5/Right-for-the-Wrong-Reasons:-Diagnosing-Syntactic-Heuristics-in-Natural-Language-Inference/graph https://www.connectedpapers.com/main/42ed4a9994e6121a9f325f5...