https://github.com/falcowinkler/nlp-getting-started
study group code for the kaggle challenge
https://github.com/falcowinkler/nlp-getting-started
Last synced: 8 months ago
JSON representation
study group code for the kaggle challenge
- Host: GitHub
- URL: https://github.com/falcowinkler/nlp-getting-started
- Owner: falcowinkler
- Created: 2020-06-11T08:44:58.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2023-02-16T01:41:11.000Z (over 2 years ago)
- Last Synced: 2025-01-18T19:08:49.481Z (9 months ago)
- Language: Python
- Size: 211 KB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 7
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
Binary tweet classification
This was an experiment to use word vectors for a simple classification task.
Idea:
- Map every word to it's corresponding word vector (we use globe embeddings)
- Feed the embeddings into a bidirectional LSTM
- Compute output by concatenating the architecture with a dense layer
Validation score on kaggle is around .7 ~ .78
Credits for the hyperparameters go to https://www.kaggle.com/shahules/basic-eda-cleaning-and-glove
Credits for the code to load glove embeddings: https://github.com/ciwin/Intent_Classification
### Run
```bash
export KAGGLE_USERNAME=
export KAGGLE_KEY=
python main.py
```