LLM-Inference/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-08-12 22:31:11 +04:00
Code Issues Packages Projects Releases Wiki Activity

model : disallow integer dflash sliding_window_pattern (#26900)

Browse Source 
* fix sliding_window_pattern

* disallow integer pattern
This commit is contained in:
Sigbjørn Skjæret
2026-08-12 14:24:10 +03:00
committed by GitHub
parent af05a42a7c
commit ece98b87f7
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/models/dflash.cpp
+1 -1
View File
@@ -66,7 +66,7 @@ void llama_model_dflash::load_arch_hparams(llama_model_loader & ml) {
// DFlash has a single rope, so the SWA rope == main rope.
if (ml.get_key(LLM_KV_ATTENTION_SLIDING_WINDOW, hparams.n_swa, false) && hparams.n_swa > 0) {
hparams.swa_type = LLAMA_SWA_TYPE_STANDARD;
ml.get_key_or_arr(LLM_KV_ATTENTION_SLIDING_WINDOW_PATTERN, hparams.is_swa_impl, hparams.n_layer());
ml.get_arr(LLM_KV_ATTENTION_SLIDING_WINDOW_PATTERN, hparams.is_swa_impl);
hparams.rope_freq_base_train_swa = hparams.rope_freq_base_train;
hparams.rope_freq_scale_train_swa = hparams.rope_freq_scale_train;
}