selection of encodings for pose and latent interpolation
- Dominant language
- Python
- Stars
- 17
- Forks
- 12
- PR merge metrics
- No merged PRs in 30d
Description
In vis.py I have noticed that we have changed the way we select latents for specific class
https://github.com/alan-turing-institute/affinity-vae/blame/17e65f90445e69b12fd81d50736f75783a120968/avae/vis.py#L1144
we seem to be selecting multiple encodings and then take the mean of them. Is it not better to just take one encoding for the purpose of interpolation and visualisation ?
Contributor guide
No contributing guide indexed for this repository
Research direction
Read avae/vis.py around line 1144 and the linked commit history first. Trace how encodings are selected for class-specific pose and latent interpolation, then establish whether the proposed single-encoding behavior is required and verify the visualization uses the agreed selection.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- data-visualization, machine-learning
- Issue type
- Feature
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100