Skip to content

v3.5.3
Compare
Choose a tag to compare
@vpirogov vpirogov released this 02 Aug 22:26
· 21 commits to rls-v3.5 since this release
v3.5.3
66f0cb9

This is a patch release containing the following changes to v3.5.2:

  • Fixed correctness issue in convolution weight gradient for small shapes on Intel GPUs (49eee6a, 281dd3b)
  • Extended MLP patterns supported by experimental Graph Compiler to cover cases relevant to ChatGLM model (ff680fc)
  • Fixed performance regression in bf16 depthwise convolution on Intel CPUs (d6c216a)
Assets 2
Loading