Added Common GPU Workflows in Docs #1980
Conversation
Codecov Report
@@ Coverage Diff @@
## master #1980 +/- ##
=======================================
Coverage 87.94% 87.94%
=======================================
Files 19 19
Lines 1485 1485
=======================================
Hits 1306 1306
Misses 179 179 Continue to review full report at Codecov.
|
|
Thanks @ToucheSir for the reviews! I only left the first one pending as I wasn't sure how to add an example of the |
Co-authored-by: Carlo Lucibello <[email protected]>
| ```julia | ||
| train_loader = Flux.DataLoader((xtrain, ytrain), batchsize = 64, shuffle = true) | ||
| # ... model, optimizer and loss definitions | ||
| for epoch in 1:nepochs | ||
| for (xtrain_batch, ytrain_batch) in train_loader | ||
| x, y = gpu(xtrain_batch), gpu(ytrain_batch) | ||
| gradients = gradient(() -> loss(x, y), parameters) | ||
| Flux.Optimise.update!(optimizer, parameters, gradients) | ||
| end | ||
| end | ||
| ``` |
There was a problem hiding this comment.
I don't believe this and subsequent code blocks need to be indented?
There was a problem hiding this comment.
I just wanted to make sure they would be in the same indentation level as the text in the sub items after rendering. But this is not mandatory at all.
Co-authored-by: Brian Chen <[email protected]>
|
Thanks @ToucheSir, @CarloLucibello and @darsnack for the reviews, comments and suggestions! |
Following the discussion in #1974, this is a very rough draft of the "Common GPU Workflows", feel free to suggest any changes or additions to this new section in the docs.
I also took the liberty of reformatting the
docs/make.jlto make the pages and titles a bit easier to read.