Replies: 3 comments
-
|
I've switched the backend back to THArray.jl, as I'm still struggling to conserve the GPU memory with Reactant.jl and optimize its speed. |
Beta Was this translation helpful? Give feedback.
-
|
It's gratifying to note that the previous installation issues with THArray.jl were merely due to its CMakeList file not being robustly written. I have since rectified this by enabling it to automatically locate the latest CUDA path, which has allowed for successful installation. The installation process is now quite straightforward, and I have successfully replicated it across multiple machines. |
Beta Was this translation helpful? Give feedback.
-
|
The sole requirement is that the CUDA version must be 12.1 or higher. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
@MilesCranmer Hi, do you have any ready-made scripts for testing SymbolicRegression.jl on SRBench or other sr benchmarks? If so, I think I could save some time.
Beta Was this translation helpful? Give feedback.
All reactions