Skip to content
Merged
Changes from all commits
Commits
Show all changes
26 commits
Select commit Hold shift + click to select a range
fa44de7
models : add llm_build_delta_net_base
ggerganov Feb 13, 2026
403e78e
cont : keep qwen35 and qwen35moe graphs intact
ggerganov Feb 14, 2026
2371dfb
cont : add comments [no ci]
ggerganov Feb 14, 2026
cff8f60
add kimi linear to delta-net-base
ymcki Feb 16, 2026
b0594c9
sync to b8057
ymcki Feb 16, 2026
6c765eb
removed unnecessary ggml_cont from g_exp_t
ymcki Feb 17, 2026
a93bcc4
removed ggml_cont from g_diff_exp_t. moved ggml_cont for o to kimi-li…
ymcki Feb 17, 2026
7b26805
removed unnecessary diag mask
ymcki Feb 17, 2026
4a6393e
cont : simplify
ggerganov Feb 17, 2026
c07977a
cont : avoid graph splits
ggerganov Feb 17, 2026
117763e
scale q after mul instead of beginning
ymcki Feb 17, 2026
6dad437
Merge branch 'ggml-org:master' into dn
ymcki Feb 17, 2026
df269dc
Merge branch 'master' of github.com:ymcki/llama.cpp into dn
ymcki Feb 17, 2026
1cea24d
scale q after mul instead of beginning
ymcki Feb 17, 2026
a6fa6d5
scale q after mul instead of beginning
ymcki Feb 17, 2026
6432f95
identical ppl
ymcki Feb 18, 2026
de6a842
cont : fix scale and decay mask
ggerganov Feb 18, 2026
23cccea
minor : remove TODO
ggerganov Feb 19, 2026
18d7b2c
block implementation for kda
ymcki Feb 23, 2026
09f0baf
block implementation for kda
ymcki Feb 23, 2026
ac46b38
remove space at the end of line 101
ymcki Feb 23, 2026
ec25a26
concat+pad
ymcki Feb 24, 2026
8c96f82
pad+binary row concat
ymcki Mar 1, 2026
aa5b816
chunk size 16 for kda
ymcki Mar 2, 2026
099645a
Merge branch 'ggml-org:master' into dn
ymcki Mar 2, 2026
f90c585
removed minor differences to master
ymcki Mar 2, 2026
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 1 addition & 3 deletions src/models/delta-net-base.cpp
Original file line number Diff line number Diff line change
@@ -1,7 +1,5 @@
#include "models.h"

#define CHUNK_SIZE 64

// utility to get one slice from the third dimension
// input dim: [x, y, c, b]
// output dim: [x, y, 1, b]
Expand Down Expand Up @@ -57,7 +55,7 @@ std::pair<ggml_tensor *, ggml_tensor *> llm_build_delta_net_base::build_delta_ne
g = ggml_permute(ctx0, g, 0, 2, 1, 3); // [g_0, n_tokens, H_v, n_seqs]
b = ggml_permute(ctx0, b, 0, 2, 1, 3); // [ 1, n_tokens, H_v, n_seqs]

const int CS = CHUNK_SIZE;
const int CS = kda ? 16 : 64; // chunk size
Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Update the PR to change only this line - the rest of the changes are not really needed.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The other change I made was removing the line "#define CHUNK_SIZE 64". Why is it still needed when no one is referencing it any more?

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, keep it. I was referring to the rest of the changes in the graph implementation.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should be ok now.


const int pad = (CS - n_tokens % CS) % CS;
const int n_chunks = (n_tokens + pad) / CS;
Expand Down
Loading