vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-05-10 04:00:53 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

models : deduplicate delta-net graphs for Qwen family (#19597)

Browse source 
* models : add llm_build_delta_net_base

* cont : keep qwen35 and qwen35moe graphs intact

* cont : add comments
This commit is contained in:
Georgi Gerganov 2026-02-16 14:35:04 +02:00 • committed by GitHub
parent d5dfc33027
commit cc45f2ada6
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
16 changed files with 428 additions and 392 deletions
Download patch file Download diff file Expand all files Collapse all files

4
src/models/kimi-linear.cpp
View file

@ -1,6 +1,8 @@
#include "models.h"
#include "ggml.h"
#include "llama-memory-recurrent.h"
#define CHUNK_SIZE 64
// Causal Conv1d function for Q,K,V
@ -65,7 +67,7 @@ static ggml_tensor * causal_conv1d(ggml_cgraph * gf, ggml_context * ctx0, ggml_t
}
llm_build_kimi_linear::llm_build_kimi_linear(const llama_model & model, const llm_graph_params & params) :
llm_graph_context_mamba(params), model(model) {
llm_build_mamba_base(params), model(model) {
ggml_tensor * cur;
ggml_tensor * inpL;