Using convolutional kernels with sizes that are multiples of 8
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 10
- Forks
- 6
- PR merge metrics
- No merged PRs in 30d
Description
DeepClean at present doesn't enjoy as much acceleration from the use of mixed precision as you might expect, and my suspicion is that because the convolutional kernels have length 7, they don't actually get executed using tensor cores on GPUs like the V100. It would be interesting to reformulate DeepClean with convolutional kernels whose lengths are multiples of 8 and see how that affects the mixed precision acceleration.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by locating DeepClean’s convolutional model implementation and the existing mixed-precision or GPU performance evaluation. Compare the current length-7 kernels with a formulation using lengths that are multiples of 8 on a V100. Done means documenting whether the reformulation improves mixed-precision acceleration without changing the model’s intended behavior.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning, performance
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100