djl icon indicating copy to clipboard operation
djl copied to clipboard

Please support CUDA12.6 as soon as possible

Open hongfeiGe opened this issue 9 months ago • 1 comments

Description

ai.djl.pytorch.jni.LibUtils : No matching cuda flavor for win-x86_64 found: cu126.

References

  • list reference and related literature
  • list known implementations

hongfeiGe avatar Feb 27 '25 07:02 hongfeiGe

@hongfeiGe

cuda12.6 should just work. DJL will download cu124/win-x86_64/native/lib/* automatically.

Can you share your pom.xml or build.gradle file? Did you set PYTORCH_PRECXX11 env var?

frankfliu avatar Mar 02 '25 18:03 frankfliu