successor commited on
Commit
d260c87
1 Parent(s): 4c2e3cf

Delete qrl-yt-metadata.jsonl

Browse files
Files changed (1) hide show
  1. qrl-yt-metadata.jsonl +0 -2
qrl-yt-metadata.jsonl DELETED
@@ -1,2 +0,0 @@
1
- {"Video ID": "Jk1YP4Y_U_0", "Channel ID": "UCv83tO5cePwHMt1952IVVHw", "Title": "Stoic Philosophy Text Generation with TensorFlow", "Time Created": "2020-04-19 11:33:45 UTC", "Time Published": "2020-04-19 13:52:43 UTC", "Duration": "1859 seconds", "Description": "Explanation of key parts to a RNN text generator built in TensorFlow with Python.\n\n\ud83e\udd16 70% Discount on the NLP With Transformers in Python course:\nhttps://bit.ly/3DFvvY5\n\nI've written a couple of Medium articles on this project, if you're interested check them out here:\nStoic Philosophy - Built by Algorithms\nhttps://towardsdatascience.com/stoic-philosophy-built-by-algorithms-9cff7b91dcbd\nSupercharged Prediction with Ensemble Learning\nhttps://towardsdatascience.com/recurrent-ensemble-learning-caffdcd94092\n\nMusic used by Lakey Inspired.\n1 - Blue Boi\n2 - Falling\nhttps://www.youtube.com/channel/UCOmy8wuTpC95lefU5d1dt2Q", "Category": "People & Blogs", "Like Count": 10.0, "Dislike Count": 0.0}
2
- {"Video ID": "gXqHd6-NKBo", "Channel ID": "UCv83tO5cePwHMt1952IVVHw", "Title": "How to Build TensorFlow Pipelines with tf.data.Dataset", "Time Created": "2020-11-02 08:23:38 UTC", "Time Published": "2020-11-02 08:57:48 UTC", "Duration": "1853 seconds", "Description": "Link to updated version (without video freeze): https://youtu.be/f6XVfgJTbp4\n\nAn introduction to building better input pipelines for Machine Learning in TF2.\n\n\ud83e\udd16 70% Discount on the NLP With Transformers in Python course:\nhttps://bit.ly/3DFvvY5\n\nLink to tf.data API docs: https://www.tensorflow.org/guide/data", "Category": "People & Blogs", "Like Count": 46.0, "Dislike Count": 9.0}