vllm kvprune wo:v1.1.0

d29c39ca · chenzk · f81ce56b · d29c39ca · d29c39ca · d29c39ca
Commit d29c39ca authored Apr 30, 2026 by chenzk
20 changed files
--- a/vllm/kvprune_legacy_save/triton_kernels/matmul_ogs_details/opt_flags_details/opt_flags_nvidia.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/matmul_ogs_details/opt_flags_details/opt_flags_nvidia.py
--- a/vllm/kvprune_legacy_save/triton_kernels/numerics.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/numerics.py
--- a/vllm/kvprune_legacy_save/triton_kernels/numerics_details/__init__.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/numerics_details/__init__.py
--- a/vllm/kvprune_legacy_save/triton_kernels/numerics_details/flexpoint.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/numerics_details/flexpoint.py
--- a/vllm/kvprune_legacy_save/triton_kernels/numerics_details/mxfp.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/numerics_details/mxfp.py
--- a/vllm/kvprune_legacy_save/triton_kernels/numerics_details/mxfp_details/__init__.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/numerics_details/mxfp_details/__init__.py
--- a/vllm/kvprune_legacy_save/triton_kernels/numerics_details/mxfp_details/_downcast_to_mxfp.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/numerics_details/mxfp_details/_downcast_to_mxfp.py
--- a/vllm/kvprune_legacy_save/triton_kernels/numerics_details/mxfp_details/_upcast_from_mxfp.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/numerics_details/mxfp_details/_upcast_from_mxfp.py
--- a/vllm/kvprune_legacy_save/triton_kernels/proton_opts.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/proton_opts.py
--- a/vllm/kvprune_legacy_save/triton_kernels/reduction_details/__init__.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/reduction_details/__init__.py
--- a/vllm/kvprune_legacy_save/triton_kernels/reduction_details/reduce_bitmatrix.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/reduction_details/reduce_bitmatrix.py
--- a/vllm/kvprune_legacy_save/triton_kernels/routing.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/routing.py
--- a/vllm/kvprune_legacy_save/triton_kernels/routing_details/__init__.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/routing_details/__init__.py
--- a/vllm/kvprune_legacy_save/triton_kernels/routing_details/_expt_data.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/routing_details/_expt_data.py
--- a/vllm/kvprune_legacy_save/triton_kernels/routing_details/_routing_compute.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/routing_details/_routing_compute.py
--- a/vllm/kvprune_legacy_save/triton_kernels/specialize.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/specialize.py
--- a/vllm/kvprune_legacy_save/triton_kernels/swiglu.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/swiglu.py
--- a/vllm/kvprune_legacy_save/triton_kernels/swiglu_details/__init__.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/swiglu_details/__init__.py
--- a/vllm/kvprune_legacy_save/triton_kernels/swiglu_details/_swiglu.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/swiglu_details/_swiglu.py
--- a/vllm/kvprune_legacy_save/triton_kernels/target_info.py
+++ b/vllm/kvprune_legacy_save/triton_kernels/target_info.py