In this project, we will generate our own Seinfeld TV scripts using RNNs. We be using part of the Seinfeld dataset of scripts from 9 seasons. The Neural Network we will build will generate a new ,"fake" TV script, based on patterns it recognizes in this training data.
- Explore the Data
- Implement Pre-processing Functions
- Build and Train the RNN
- Hyperparameters Tuning
- Generate TV Scripts
- Epoch: 1/20 Loss: 5.299629842758178
- Epoch: 20/20 Loss: 3.2879079582691193
- Script Sample:
jerry:...
jerry: i don't think so.
elaine: i thought i would have a good time, i know. it's a misprint.
kramer: well, it's a big mistake!
hoyt: so you have no idea what the problem is in the bubble.
jerry: what?
george: i know.
elaine: you know, you have a good idea.
george: oh, i can't do it.
george: you want to get a job?
elaine: i can't.
jerry: i can't do this!
elaine: what?
elaine: i know, i don't have it to get to know.
hoyt: i thought he liked me.
jerry: what?
jerry: you don't know how the new woman has ever done that. i don't know why you don't want to talk with this guy, and i was wondering if it was an accident.
george: what? what is that?
george: oh, it's my fault.
jerry: you know, it's not a problem.
jerry: you want to be a liar.
elaine: oh, yeah!
jerry: what?
jerry: no, it's a very important time.
jerry: i don't know.
elaine: you can't get the hell out of my office.
jerry: what are you doing?
jerry: no, i don't know how you're going to be.
jerry: what is the problem with you?
kramer: oh, no, no. no. no, no. no. no. no, no. it's a little rough.
george: i know.
jerry: i don't know. it's just a problem. i don't want to get a piece of cake!
hoyt: i have to get the feeling of it.
george: what?
jerry: i think it was a little messy.
elaine: oh, no. no, no. no- no. no,
Visit the detailed Jupyter Notebook: TV_Script_Generation.ipynb