meetps / meetps/tf-3dgan

testGAN function for generating new objects

Open
#21 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
287
Forks
78
PR merge metrics
No merged PRs in 30d

Description

I'm using the 3dgan_mit_biasfree model. I want to generate new instances.
first of all, I set changed one of the lines to this::
net_g_test = generator(z_vector, phase_train=False, reuse=False)
because it is testing so phase_train should not be true, right? Also, as another person said in another issue, there seems to be a problem with batchnorm when setting reuse=True.

Also, around line 250, there is next_sigma, which is the SD of the normal distributon from which elements of the latent vector z are drawn. My question is since I hardcoded it as .33 [what you found to be good during training], instead of raw_input(), is there a better value I should be using?

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the testGAN function and the code around line 250, then trace how phase_train, reuse, batch normalization, and next_sigma are passed into generation. Done means validating the inference settings and latent-vector standard deviation needed to generate new instances, with a reproducible test or example.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, tensorflow
Domain
machine-learning
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
30/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.