From 221edb9ef1aa2155404d0a121921f7e0e418feda Mon Sep 17 00:00:00 2001 From: Andrei Betlen Date: Sat, 24 Feb 2024 23:47:29 -0500 Subject: [PATCH] feat: Update llama.cpp --- llama_cpp/llama_cpp.py | 4 ++++ vendor/llama.cpp | 2 +- 2 files changed, 5 insertions(+), 1 deletion(-) diff --git a/llama_cpp/llama_cpp.py b/llama_cpp/llama_cpp.py index f4d523b..b8e74d7 100644 --- a/llama_cpp/llama_cpp.py +++ b/llama_cpp/llama_cpp.py @@ -245,6 +245,8 @@ LLAMA_TOKEN_TYPE_BYTE = 6 # 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_MOSTLY_IQ3_S = 26, // except 1d tensors +# LLAMA_FTYPE_MOSTLY_IQ3_M = 27, // except 1d tensors # LLAMA_FTYPE_GUESSED = 1024, // not specified in the model file # }; @@ -272,6 +274,8 @@ 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_MOSTLY_IQ3_S = 26 +LLAMA_FTYPE_MOSTLY_IQ3_M = 27 LLAMA_FTYPE_GUESSED = 1024 # enum llama_rope_scaling_type { diff --git a/vendor/llama.cpp b/vendor/llama.cpp index 15499eb..9e359a4 160000 --- a/vendor/llama.cpp +++ b/vendor/llama.cpp @@ -1 +1 @@ -Subproject commit 15499eb94227401bdc8875da6eb85c15d37068f7 +Subproject commit 9e359a4f47c1b2dceb99e29706c9f7403d32ab5e