From: Victor Villar <redacted>
Date: Fri, 20 Mar 2026 14:16:09 +0000 (+0100)
Subject: model : fix Granite Hybrid type check for 7B.A1B (#20795)
X-Git-Tag: upstream/0.0.8611~156
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=58c81f7e811d82b9ba62002f40d7af56134b61fc;p=pkg%2Fggml%2Fsources%2Fllama.cpp

model : fix Granite Hybrid type check for 7B.A1B (#20795)

* Check granite hybriid expert count to set type as LLM_TYPE_7B_A1B or LLM_TYPE_1B

* Use feed fwd dim instead of num of experts

Co-authored-by: Sigbjørn Skjæret <redacted>

---------

Co-authored-by: Sigbjørn Skjæret <redacted>
---

diff --git a/src/llama-model.cpp b/src/llama-model.cpp
index 06e064535..f8caad288 100644
--- a/src/llama-model.cpp
+++ b/src/llama-model.cpp
@@ -2058,7 +2058,7 @@ void llama_model::load_hparams(llama_model_loader & ml) {
 
                 switch (hparams.n_embd) {
                     case 768: type = LLM_TYPE_350M; break;
-                    case 1536: type = (hparams.n_embd == 2048 ? LLM_TYPE_7B_A1B : LLM_TYPE_1B); break;
+                    case 1536: type = (hparams.n_ff() == 512 ? LLM_TYPE_7B_A1B : LLM_TYPE_1B); break;
                     case 2048: case 2560: type = LLM_TYPE_3B; break;
                     case 4096: type = LLM_TYPE_32B; break;
                     default: type = LLM_TYPE_UNKNOWN;