millaguie/llama.cpp
1
0 Fork 0
Code Issues Pull Requests Actions 30 Packages Projects Releases Wiki Activity

graph : ensure DS32 kq_mask_lid is F32 (#23864)

Browse Source
This commit is contained in:
Sigbjørn Skjæret authored and GitHub committed 2026-05-29 19:55:14 +02:00
1 parent b5f52280fb
commit 764f1e64a1
2 files changed
+11 -7
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

src/llama-graph.h
+4 -4
View File
@@ -399,10 +399,10 @@ public:
ggml_tensor * self_k_idxs_mla = nullptr; // I64 [n_batch]
ggml_tensor * self_k_idxs_lid = nullptr; // I64 [n_batch]
ggml_tensor * self_kq_mask_mla = nullptr; // F32 [n_kv, n_batch/n_stream, 1, n_stream]
ggml_tensor * self_kq_mask_mla_cnv = nullptr; // [n_kv, n_batch/n_stream, 1, n_stream]
ggml_tensor * self_kq_mask_lid = nullptr; // F32 [n_kv, n_batch/n_stream, 1, n_stream]
ggml_tensor * self_kq_mask_lid_cnv = nullptr; // [n_kv, n_batch/n_stream, 1, n_stream]
ggml_tensor * self_kq_mask_mla = nullptr; // F32/F16 [n_kv, n_batch/n_stream, 1, n_stream]
ggml_tensor * self_kq_mask_mla_cnv = nullptr; // [n_kv, n_batch/n_stream, 1, n_stream]
ggml_tensor * self_kq_mask_lid = nullptr; // F32 [n_kv, n_batch/n_stream, 1, n_stream]
ggml_tensor * self_kq_mask_lid_cnv = nullptr; // [n_kv, n_batch/n_stream, 1, n_stream]
ggml_tensor * self_k_rot_lid = nullptr;