feat fused_adamw #938
Merged
feat fused_adamw #938
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
lljbash
reviewed
Sep 25, 2024
Comment on lines
1330
to
1361
| std::vector<diopiTensorHandle_t> diopiTensorHandles_self(self.size()); | ||
| for(size_t i=0; i < self.size(); ++i){ | ||
| diopiConstTensorHandle_t const_handle = dipu::diopi_helper::toDiopiTensorHandle(self.at(i)); | ||
| diopiTensorHandle_t handle = const_cast<diopiTensorHandle_t>(const_handle); | ||
| diopiTensorHandles_self[i] = handle; | ||
| } | ||
| std::vector<diopiConstTensorHandle_t> diopiTensorHandles_grads(grads.size()); | ||
| for(size_t i=0; i < grads.size(); ++i){ | ||
| diopiTensorHandles_grads[i] = dipu::diopi_helper::toDiopiTensorHandle(grads.at(i)); | ||
| } | ||
| std::vector<diopiTensorHandle_t> diopiTensorHandles_exp_avgs(exp_avgs.size()); | ||
| for(size_t i=0; i < exp_avgs.size(); ++i){ | ||
| diopiConstTensorHandle_t const_handle = dipu::diopi_helper::toDiopiTensorHandle(exp_avgs.at(i)); | ||
| diopiTensorHandle_t handle = const_cast<diopiTensorHandle_t>(const_handle); | ||
| diopiTensorHandles_exp_avgs[i] = handle; | ||
| } | ||
| std::vector<diopiTensorHandle_t> diopiTensorHandles_exp_avg_sqs(exp_avg_sqs.size()); | ||
| for(size_t i=0; i < exp_avg_sqs.size(); ++i){ | ||
| diopiConstTensorHandle_t const_handle = dipu::diopi_helper::toDiopiTensorHandle(exp_avg_sqs.at(i)); | ||
| diopiTensorHandle_t handle = const_cast<diopiTensorHandle_t>(const_handle); | ||
| diopiTensorHandles_exp_avg_sqs[i] = handle; | ||
| } | ||
| std::vector<diopiTensorHandle_t> diopiTensorHandles_max_exp_avg_sqs(max_exp_avg_sqs.size()); | ||
| for(size_t i=0; i < max_exp_avg_sqs.size(); ++i){ | ||
| diopiConstTensorHandle_t const_handle = dipu::diopi_helper::toDiopiTensorHandle(max_exp_avg_sqs.at(i)); | ||
| diopiTensorHandle_t handle = const_cast<diopiTensorHandle_t>(const_handle); | ||
| diopiTensorHandles_max_exp_avg_sqs[i] = handle; | ||
| } | ||
| std::vector<diopiConstTensorHandle_t> diopiTensorHandles_state_steps(state_steps.size(), nullptr); | ||
| for(size_t i=0; i < state_steps.size(); ++i){ | ||
| diopiTensorHandles_state_steps[i] = dipu::diopi_helper::toDiopiTensorHandle(state_steps.at(i)); | ||
| } |
There was a problem hiding this comment.
削减重复代码，考虑：
- 提取函数
- 使用 std::transform
There was a problem hiding this comment.
toDiopiTensorHandleVector
已经有这个函数了
| ), | ||
| ) | ||
|
|
||
| @onlyOn("CUDA") |
There was a problem hiding this comment.
如果期望所有芯片都实现，应该用 skipon 比较好
lljbash
approved these changes
Oct 11, 2024
fandaoyi
approved these changes
Oct 11, 2024
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.
No description provided.