mikhail/llama.cpp-mtp-turboquant
1
0
Fork 0
Code Issues Pull Requests Actions 35 Packages Projects Releases Wiki Activity

model : support NVFP4 tensors for Gemma4 (#21971)

Browse Source 
* support nvfp4 tensors for Gemma4

* add wo_s to build_attn

* add wo_s to build_attn

* fix glm4
This commit is contained in:
Sigbjørn Skjæret
2026-04-16 16:51:47 +02:00
committed by GitHub
parent b572d1ecd6
commit f772f6e434
105 changed files with 149 additions and 148 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/models/gemma3n-iswa.cpp
+2 -2
View File
@@ -103,7 +103,7 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
cb(Kcur, "Kcur_pos", il);
cur = build_attn(inp_attn, model.layers[il].wo,
NULL, Qcur, Kcur, Vcur, nullptr, nullptr, nullptr,
NULL, model.layers[il].wo_s, Qcur, Kcur, Vcur, nullptr, nullptr, nullptr,
hparams.f_attention_scale, il);
} else {
// reuse KV cache of earlier layers
@@ -119,7 +119,7 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
cb(Qcur, "Qcur_pos", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, nullptr, nullptr, nullptr, nullptr, nullptr, hparams.f_attention_scale, il);
}
cur = build_norm(cur, model.layers[il].attn_post_norm, NULL, LLM_NORM_RMS, il);