feat: Update llama.cpp

abetlen · abetlen · commit 4edde21b3d68 · 2024-02-21T11:05:58.000-05:00
diff --git a/llama_cpp/llama_cpp.py b/llama_cpp/llama_cpp.py
@@ -191,6 +191,7 @@ def _load_shared_library(lib_base_name: str):
 #     LLAMA_FTYPE_MOSTLY_Q3_K_XS       = 22, // except 1d tensors
 #     LLAMA_FTYPE_MOSTLY_IQ3_XXS       = 23, // except 1d tensors
 #     LLAMA_FTYPE_MOSTLY_IQ1_S         = 24, // except 1d tensors
+#     LLAMA_FTYPE_MOSTLY_IQ4_NL        = 25, // except 1d tensors
 
 #     LLAMA_FTYPE_GUESSED = 1024, // not specified in the model file
 # };
@@ -217,6 +218,7 @@ def _load_shared_library(lib_base_name: str):
 LLAMA_FTYPE_MOSTLY_Q3_K_XS = 22
 LLAMA_FTYPE_MOSTLY_IQ3_XXS = 23
 LLAMA_FTYPE_MOSTLY_IQ1_S = 24
+LLAMA_FTYPE_MOSTLY_IQ4_NL = 25
 LLAMA_FTYPE_GUESSED = 1024
 
 # enum llama_rope_scaling_type {
diff --git a/vendor/llama.cpp b/vendor/llama.cpp
@@ -1 +1 @@
-Subproject commit c0a8c6db371cb3e4379900867b948879f5842201
+Subproject commit 89febfed9322c8849520dc63c93ee4f5fd72556e
