From: SamareshSingh <redacted>
Date: Sat, 4 Apr 2026 11:05:10 +0000 (-0500)
Subject: llama-model: read final_logit_softcapping for Gemma 4 (#21390)
X-Git-Tag: upstream/0.0.8681~19
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=650bf14eb9a922de0f88c9523a271159cc5ae469;p=pkg%2Fggml%2Fsources%2Fllama.cpp

llama-model: read final_logit_softcapping for Gemma 4 (#21390)
---

diff --git a/src/llama-model.cpp b/src/llama-model.cpp
index ba935340f..79d08ff41 100644
--- a/src/llama-model.cpp
+++ b/src/llama-model.cpp
@@ -1279,6 +1279,7 @@ void llama_model::load_hparams(llama_model_loader & ml) {
                 ml.get_key(LLM_KV_EMBEDDING_LENGTH_PER_LAYER,  hparams.n_embd_per_layer);
                 ml.get_key(LLM_KV_ATTENTION_KEY_LENGTH_SWA,    hparams.n_embd_head_k_swa);
                 ml.get_key(LLM_KV_ATTENTION_VALUE_LENGTH_SWA,  hparams.n_embd_head_v_swa);
+                ml.get_key(LLM_KV_FINAL_LOGIT_SOFTCAPPING,     hparams.f_final_logit_softcapping, false);
 
                 switch (hparams.n_layer) {
                     case 35: type = LLM_TYPE_E2B; break;