ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-04-27 01:49:28 +00:00
Code Issues Packages Projects Releases Wiki Activity

GLM-4.7-Flash support (#1168)

Browse Source 
* GLM-4.7-Flash support

* Model type

* Make FA work for mla != 0
This commit is contained in:
Kawrakow
2026-01-20 12:46:52 +02:00
committed by GitHub
parent ef5f17940c
commit 132a01d25d
3 changed files with 22 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
src/llama-hparams.cpp
View File

@@ -750,7 +750,7 @@ void llm_load_hparams(
{
if (hparams.n_head_kv() == 1) {
int n_nead_kv = hparams.n_gqa();
if (n_nead_kv%16 != 0 || hparams.n_embd_head_k != 576 || hparams.n_embd_head_v != 512 ||
if (n_nead_kv%4 != 0 || hparams.n_embd_head_k != 576 || hparams.n_embd_head_v != 512 ||
hparams.n_rot != 64) {
printf("==========================================================================\n");
printf("Detected incompatible DeepSeek model without a known way to fixc it.\n");
@@ -788,6 +788,7 @@ void llm_load_hparams(
switch (hparams.n_layer) {
case 27: model.type = e_model::MODEL_16B; break;
case 47: model.type = e_model::MODEL_30B_A3B; break; // GLM-4.7-Flash
case 60: model.type = e_model::MODEL_236B; break;
case 61: model.type = e_model::MODEL_671B; break;
default: model.type = e_model::MODEL_UNKNOWN;