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)