ClaudeBot/llama.cpp
0
0
Fork 1
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-09-17 20:31:47 +02:00
Code Issues Packages Projects Releases Wiki Activity

model : fix MTP context kv cache allocation for deepseek2, glm4moe, … (#28630)

Browse Source 
* model : fix MTP context kv cache allocation for deepseek2, glm4moe, cohere2moe architectures (#28626)

* model: add inverse architecture gating and comprehensive architecture testing for mtp layer filtering

* model : slim NextN filter comment, drop test-llama-archs changes
This commit is contained in:
Logan Chu
2026-09-11 12:02:31 +03:00
committed by GitHub
parent b0dcb8192b
commit 5cdd3d1dad
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/llama-model.cpp
+3 -3
View File
@@ -2644,9 +2644,9 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
filter = [&](uint32_t il) { return il >= hparams.n_layer(); };
}
if ((arch == LLM_ARCH_STEP35 || arch == LLM_ARCH_HY_V3 || arch == LLM_ARCH_GLM_DSA ||
arch == LLM_ARCH_MIMO2 || arch == LLM_ARCH_DEEPSEEK32) &&
hparams.n_layer_nextn > 0) {
// don't filter when n_layer_nextn is repurposed for a router layer the trunk attends
// or when a model is entirely n_layer_nextn layers and has no trunk
if (hparams.n_layer_nextn > 0 && hparams.n_layer() > 0 && hparams.router_layer < 0) {
if (params.ctx_type == LLAMA_CONTEXT_TYPE_MTP) {
filter = [&](uint32_t il) { return il >= hparams.n_layer(); };
} else {