Skip to content

Conversation

ahirner
Copy link

@ahirner ahirner commented Apr 15, 2017

Function generate_sequence() resolves #12 and takes an initial string, a dictionary that overrides hidden states, the stopping condition, and a mode for which (greedy) sampler to use.
Usage:
positive = mdl.generate_sequence(base_str, override={2388 : 1.0}, len_add = '..', sampling = 1)
See docstring for further explanation.

The options are tested in test_generative.py. Sampling two sentences after I couldn’t figure out with a fixed positive sentiment (2388 : 1.0) and stochastic sampling after separate words yields pretty convincing examples:
I couldn’t figure out what all the complaints are about but I love it and I can’t wait to see what else she makes in the future. I have not stopped laughing out loud since the first book of hers I have read.

Example with a negative sentiment (2388 : -1.5):
I couldn’t figure out how to get the product to work with some trial version. Not even a 2 page document with a text appeared on the left hand side correctly.

@ahirner ahirner changed the title Added generation (resolves #12) Added generation Apr 15, 2017
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Generation

1 participant