kompute : add gemma, phi-2, qwen2, and stablelm to whitelist

Signed-off-by: Jared Van Bortel <[email protected]>
nomic-ai · Feb 21, 2024 · 7d4ced8 · 7d4ced8
1 parent bf31654
commit 7d4ced8
Showing 1 changed file with 2 additions and 1 deletion.
diff --git a/llama.cpp b/llama.cpp
@@ -4553,7 +4553,8 @@ static int llama_model_load(const std::string & fname, llama_model & model, llam
 
 #ifdef GGML_USE_KOMPUTE
         if (params.n_gpu_layers > 0 && (
-            !(model.arch == LLM_ARCH_LLAMA || model.arch == LLM_ARCH_FALCON)
+            !(model.arch == LLM_ARCH_FALCON || model.arch == LLM_ARCH_GEMMA || model.arch == LLM_ARCH_LLAMA ||
+              model.arch == LLM_ARCH_PHI2   || model.arch == LLM_ARCH_QWEN2 || model.arch == LLM_ARCH_STABLELM)
             || model.hparams.n_expert > 0
             || !(
                 model.ftype == LLAMA_FTYPE_ALL_F32 ||