Implement cuDNN activation functions. #33
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- Add wrapper functions for `cudnnActivationForward` and `cudnnActivationBackward`. - Move `relu` to be backend-defined. - Implement `relu` and `relu_grad` using cuDNN, add tests. - Minor: change out-of-bounds indexing via `Dimensions.apply` to clearly throw `IndexOutOfBoundsException`. TODO: - Support `relu` for non-4D tensors. This will likely require shape padding with "1" dimensions to work with the cuDNN API. - Implement other activation functions.
dan-zheng
commented
Oct 21, 2018
| } | ||
| } | ||
|
|
||
| override def tanh(x: Tensor) = x.map(s => Math.tanh(s).toFloat) |
There was a problem hiding this comment.
It'd be good to decouple these backend functions from Tensor methods (map and add_oneMinusSquare_mult).
There was a problem hiding this comment.
yes, I agree. We are sort things out as we encounter them.
dan-zheng
changed the title
I accidentally copied tanh's gradient.
dan-zheng
commented
Oct 21, 2018
| "CUDNN_CALL(cudnnActivationBackward(\n" + | ||
| " cudnnHandle, act_desc,\n" + | ||
| " ", one, ", x_desc, ", res.x.data, ", x_desc, ", res.d.data, ", x_desc, ", input.x.data, ",\n", | ||
| " ", zero, ", x_desc, ", inputGrad.data, "));\n" + |
There was a problem hiding this comment.
As suggested by @feiwang3311: rather than using beta = 0 and input.d += inputGrad, use beta = 1.
Same elsewhere (for conv2d, etc).
I'll take on this.
18 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
cudnnActivationForwardandcudnnActivationBackward.TODO:
padding with "1" dimensions to work with the cuDNN API.