Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
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.