Update README.md
Browse files
README.md
CHANGED
@@ -2,6 +2,6 @@ Creation Steps
|
|
2 |
- Downloaded [5 Million Song Dataset](https://www.kaggle.com/datasets/nikhilnayak123/5-million-song-lyrics-dataset) from Kaggle
|
3 |
- Selected quality artists, as defined by me
|
4 |
- Remove songs featuring any [profanity](https://github.com/surge-ai/profanity)
|
5 |
-
-
|
6 |
-
-
|
7 |
- Computed four sets of embeddings using all-MiniLM-L12-v2, all-distilroberta-v1, text-embedding-ada-002, and average_word_embeddings_glove.840B.300d
|
|
|
2 |
- Downloaded [5 Million Song Dataset](https://www.kaggle.com/datasets/nikhilnayak123/5-million-song-lyrics-dataset) from Kaggle
|
3 |
- Selected quality artists, as defined by me
|
4 |
- Remove songs featuring any [profanity](https://github.com/surge-ai/profanity)
|
5 |
+
- Added normalized version of lyrics (used for GloVe embedding only)
|
6 |
+
- lower case, remove punctuation, remove stopwords, lemmatize)
|
7 |
- Computed four sets of embeddings using all-MiniLM-L12-v2, all-distilroberta-v1, text-embedding-ada-002, and average_word_embeddings_glove.840B.300d
|