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

spec : add DFlash support (#22105)

Browse Source 
* spec: add DFlash v2 support

* dflash: support sliding window attention per layer_types

* docs: add dflash section

---------

Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
This commit is contained in:
Ruixiang Wang
2026-06-28 16:01:34 +03:00
committed by GitHub
co-authored by Kashif Rasul
parent c1a1c8ee94
commit d1b34251bc
14 changed files with 712 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/llama-model.cpp
+5 -1
View File
@@ -291,6 +291,8 @@ static llama_model * llama_model_mapping(llm_arch arch, const llama_model_params
return new llama_model_mistral3(params);
case LLM_ARCH_EAGLE3:
return new llama_model_eagle3(params);
case LLM_ARCH_DFLASH:
return new llama_model_dflash(params);
case LLM_ARCH_MIMO2:
return new llama_model_mimo2(params);
case LLM_ARCH_KIMI_LINEAR:
@@ -2494,6 +2496,7 @@ llama_rope_type llama_model_rope_type(const llama_model * model) {
case LLM_ARCH_STEP35:
case LLM_ARCH_TALKIE:
case LLM_ARCH_MELLUM:
case LLM_ARCH_DFLASH:
return LLAMA_ROPE_TYPE_NEOX;
case LLM_ARCH_QWEN2VL:
@@ -2617,7 +2620,8 @@ bool llama_model_has_encoder(const llama_model * model) {
switch (model->arch) {
case LLM_ARCH_T5:
case LLM_ARCH_T5ENCODER:
case LLM_ARCH_EAGLE3: return true;
case LLM_ARCH_EAGLE3:
case LLM_ARCH_DFLASH: return true;
default: return false;
}
}