- the movie poster breaks as soon as it should start predicting your score
- an option to skip movies you haven't soon would be helpful
- it showed me a movie I had already rated in its initial data gathering phase
I'm interested in playing around with this even after a rough demo
Cool. I starred the repo on github so I can follow your progress. I might have a use for this in a project of mine and I like the API you've established
I wasted time rating about 60 movies - not a large data set mind you - and it predicted I would rate Spirited Away as a 1 star. I imagine because I rated most other "Animated" and "Adventure" movies prior to it relatively poorly.
Spirited Away has been my favorite movie since I was a child. So I had a slight chuckle about the prediction. :)
That's expected behavior for how the network was trained, so it's not meant as criticism.
I'd love to play with neural networks more but lack the creativity to apply them to something interesting...and what I'm really interested in training them on is far beyond my abilities.