mirror of
https://github.com/LostRuins/koboldcpp.git
synced 2026-09-18 16:55:14 +02:00
Merge branch 'upstream' into concedo_experimental
# Conflicts: # .github/workflows/build-cuda-windows.yml # .github/workflows/release.yml # app/llama.cpp # build-xcframework.sh # docs/speculative.md # ggml/src/ggml-opencl/ggml-opencl.cpp # ggml/src/ggml-opencl/kernels/cvt.cl # ggml/src/ggml-webgpu/ggml-webgpu-shader-lib.hpp # ggml/src/ggml-webgpu/ggml-webgpu.cpp # ggml/src/ggml-webgpu/wgsl-shaders/set_rows.wgsl # requirements/requirements-server-bench.txt # scripts/server-bench.py # tests/test-backend-ops.cpp # tests/test-llama-archs.cpp # tools/cli/README.md # tools/completion/README.md # tools/server/README.md
This commit is contained in:
+7
-3
@@ -2656,14 +2656,18 @@ llm_graph_input_attn_k_dsa * llm_graph_context::build_attn_inp_k_dsa() const {
|
||||
inp->self_k_idxs_mla = mctx_cur->get_mla()->build_input_k_idxs(ctx0, ubatch);
|
||||
|
||||
inp->self_kq_mask_mla = build_attn_inp_kq_mask(ctx0, mctx_cur->get_mla(), ubatch, cparams);
|
||||
inp->self_kq_mask_mla_cnv = cparams.flash_attn ? ggml_cast(ctx0, inp->self_kq_mask_mla, GGML_TYPE_F16) : inp->self_kq_mask_mla;
|
||||
inp->self_kq_mask_mla_cnv = inp->self_kq_mask_mla;
|
||||
}
|
||||
|
||||
{
|
||||
inp->self_k_idxs_lid = mctx_cur->get_lid()->build_input_k_idxs(ctx0, ubatch);
|
||||
|
||||
inp->self_kq_mask_lid = build_attn_inp_kq_mask(ctx0, mctx_cur->get_lid(), ubatch, cparams);
|
||||
inp->self_kq_mask_lid_cnv = cparams.flash_attn ? ggml_cast(ctx0, inp->self_kq_mask_lid, GGML_TYPE_F16) : inp->self_kq_mask_lid;
|
||||
// ensure F32 mask
|
||||
auto cparams_copy = cparams;
|
||||
cparams_copy.flash_attn = false;
|
||||
|
||||
inp->self_kq_mask_lid = build_attn_inp_kq_mask(ctx0, mctx_cur->get_lid(), ubatch, cparams_copy);
|
||||
inp->self_kq_mask_lid_cnv = inp->self_kq_mask_lid;
|
||||
|
||||
inp->self_k_rot_lid = mctx_cur->get_lid()->build_input_k_rot(ctx0);
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user