Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

MNIST op modeling - relu (PoC) #1326

Open
skrsticTT opened this issue Nov 19, 2024 · 0 comments
Open

MNIST op modeling - relu (PoC) #1326

skrsticTT opened this issue Nov 19, 2024 · 0 comments

Comments

@skrsticTT
Copy link

Idea
The main goal of this task is to model relu op - in other words, we want to provide an estimation mechanism that will gave as a fast way to predict how long some relu op with the specific configuration will last (tensor size, tensor layout, data format...).

Modeling
Hypothesis is that relu op duration can be estimated with linear model. We need to measure training data that will provide us a proper fit, and than use that coefficients to estimate required op. Testing data needs to be segmented - every segment is going to have their coefficients.

Error
Also, we need to provide an expected error for this op, to make sure that Optimizer can be aware of how good this estimation is.

Some insights and results are going to be provided in the comment section.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

1 participant