baalajimaestro/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

feat: Update llama.cpp

Browse source
This commit is contained in:
Andrei Betlen 2024-02-21 11:05:58 -05:00
parent f57b01ac9b
commit 4edde21b3d
2 changed files with 3 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
llama_cpp/llama_cpp.py
View file

@ -191,6 +191,7 @@ LLAMA_TOKEN_TYPE_BYTE = 6
# LLAMA_FTYPE_MOSTLY_Q3_K_XS = 22, // except 1d tensors # LLAMA_FTYPE_MOSTLY_Q3_K_XS = 22, // except 1d tensors
# LLAMA_FTYPE_MOSTLY_IQ3_XXS = 23, // except 1d tensors # LLAMA_FTYPE_MOSTLY_IQ3_XXS = 23, // except 1d tensors
# LLAMA_FTYPE_MOSTLY_IQ1_S = 24, // 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 # LLAMA_FTYPE_GUESSED = 1024, // not specified in the model file
# }; # };
@ -217,6 +218,7 @@ LLAMA_FTYPE_MOSTLY_Q2_K_S = 21
LLAMA_FTYPE_MOSTLY_Q3_K_XS = 22 LLAMA_FTYPE_MOSTLY_Q3_K_XS = 22
LLAMA_FTYPE_MOSTLY_IQ3_XXS = 23 LLAMA_FTYPE_MOSTLY_IQ3_XXS = 23
LLAMA_FTYPE_MOSTLY_IQ1_S = 24 LLAMA_FTYPE_MOSTLY_IQ1_S = 24
LLAMA_FTYPE_MOSTLY_IQ4_NL = 25
LLAMA_FTYPE_GUESSED = 1024 LLAMA_FTYPE_GUESSED = 1024
# enum llama_rope_scaling_type { # enum llama_rope_scaling_type {

2
vendor/llama.cpp vendored

@ -1 +1 @@
Subproject commit c0a8c6db371cb3e4379900867b948879f5842201 Subproject commit 89febfed9322c8849520dc63c93ee4f5fd72556e