Intel® Software News Updates
…FlexAttention support for X86 CPUs was added through the TorchInductor CPP backend. This extends current CPP template abilities to support broad attention variants (e.g., PageAttention, which is critical for LLMs inference…