vllm kvprune:v1.0.1

f81ce56b · chenzk · 2b7160c6 · 2b7160c6 · 2b7160c6 · 2b7160c6
Commit f81ce56b authored Apr 23, 2026 by chenzk
20 changed files
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/proton_opts.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/proton_opts.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/reduction_details/__init__.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/reduction_details/__init__.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/reduction_details/reduce_bitmatrix.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/reduction_details/reduce_bitmatrix.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/routing.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/routing.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/routing_details/__init__.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/routing_details/__init__.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/routing_details/_expt_data.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/routing_details/_expt_data.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/routing_details/_routing_compute.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/routing_details/_routing_compute.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/specialize.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/specialize.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/swiglu.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/swiglu.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/swiglu_details/__init__.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/swiglu_details/__init__.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/swiglu_details/_swiglu.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/swiglu_details/_swiglu.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/target_info.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/target_info.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/__init__.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/__init__.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/layout.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/layout.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/layout_details/__init__.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/layout_details/__init__.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/layout_details/base.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/layout_details/base.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/layout_details/blackwell_scale.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/layout_details/blackwell_scale.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/layout_details/blackwell_value.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/layout_details/blackwell_value.py
--- a/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/layout_details/cdna4_scale.py
+++ b/vllm/compactor-vllm/src/compactor_vllm/triton_kernels/tensor_details/layout_details/cdna4_scale.py