Merge tag 'v0.14.0' into v0.14.0-dev

7e63ef82 · zhuwenwen · 8cbcac5d · b17039bc · 7e63ef82 · 7e63ef82
Commit 7e63ef82 authored Jan 21, 2026 by zhuwenwen
20 changed files
--- a/csrc/cpu/cpu_wna16.cpp
+++ b/csrc/cpu/cpu_wna16.cpp
--- a/csrc/cpu/dnnl_helper.cpp
+++ b/csrc/cpu/dnnl_helper.cpp
--- a/csrc/cpu/micro_gemm/cpu_micro_gemm_amx.hpp
+++ b/csrc/cpu/micro_gemm/cpu_micro_gemm_amx.hpp
--- a/csrc/cpu/micro_gemm/cpu_micro_gemm_impl.hpp
+++ b/csrc/cpu/micro_gemm/cpu_micro_gemm_impl.hpp
--- a/csrc/cpu/micro_gemm/cpu_micro_gemm_vec.hpp
+++ b/csrc/cpu/micro_gemm/cpu_micro_gemm_vec.hpp
--- a/csrc/cpu/scratchpad_manager.cpp
+++ b/csrc/cpu/scratchpad_manager.cpp
--- a/csrc/cpu/scratchpad_manager.h
+++ b/csrc/cpu/scratchpad_manager.h
--- a/csrc/cpu/torch_bindings.cpp
+++ b/csrc/cpu/torch_bindings.cpp
--- a/csrc/cpu/utils.cpp
+++ b/csrc/cpu/utils.cpp
--- a/csrc/cpu/utils.hpp
+++ b/csrc/cpu/utils.hpp
--- a/csrc/cumem_allocator.cpp
+++ b/csrc/cumem_allocator.cpp
--- a/csrc/fused_qknorm_rope_kernel.cu
+++ b/csrc/fused_qknorm_rope_kernel.cu
--- a/csrc/moe/grouped_topk_kernels.cu
+++ b/csrc/moe/grouped_topk_kernels.cu
--- a/csrc/moe/marlin_moe_wna16/.gitignore
+++ b/csrc/moe/marlin_moe_wna16/.gitignore
 sm*_kernel_*.cu
 kernel_selector.h
+kernel_*.cu
--- a/csrc/moe/marlin_moe_wna16/generate_kernels.py
+++ b/csrc/moe/marlin_moe_wna16/generate_kernels.py
--- a/csrc/moe/marlin_moe_wna16/kernel.h
+++ b/csrc/moe/marlin_moe_wna16/kernel.h
--- a/csrc/moe/marlin_moe_wna16/marlin_template.h
+++ b/csrc/moe/marlin_moe_wna16/marlin_template.h
--- a/csrc/moe/marlin_moe_wna16/ops.cu
+++ b/csrc/moe/marlin_moe_wna16/ops.cu
--- a/csrc/moe/torch_bindings.cpp
+++ b/csrc/moe/torch_bindings.cpp
--- a/csrc/ops.h
+++ b/csrc/ops.h