carpedm20 / carpedm20/DCGAN-tensorflow
Training good till epoch 20 then goes horribly bad...
- Dominant language
- JavaScript
- Stars
- 7.2k
- Forks
- 2.6k
- PR merge metrics
- No merged PRs in 30d
Description
I had these settings
CelebA
265 input
128 output
25 epochs
--crop --train
and everything was going fine till epoch 20 then things got very badly? Anybody an idea what's happening and if I can redo it from epoch 19?? Waste of amazon EC2 GPU time and money otherwise?
Contributor guide
No contributing guide indexed for this repository
Research direction
Start by reproducing the reported CelebA training setup: 265 input, 128 output, 25 epochs, with --crop --train. Compare the generated outputs around epochs 19–21 and inspect how training checkpoints are saved or resumed. Done means the cause of the degradation and a verified way to resume from epoch 19 are identified.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- tensorflow
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100