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

ggml : add TQ1_0 and TQ2_0 ternary quantization types

Browse source
This commit is contained in:
Francis Couture-Harpin 2024-07-30 17:55:54 -04:00
parent 79a278e922
commit 77b8f84ae7
10 changed files with 563 additions and 16 deletions
Download patch file Download diff file Expand all files Collapse all files

6
include/llama.h
View file

@ -166,8 +166,10 @@ extern "C" {
LLAMA_FTYPE_MOSTLY_Q4_0_4_4 = 33, // except 1d tensors
LLAMA_FTYPE_MOSTLY_Q4_0_4_8 = 34, // except 1d tensors
LLAMA_FTYPE_MOSTLY_Q4_0_8_8 = 35, // except 1d tensors
LLAMA_FTYPE_MOSTLY_Q2_2 = 36, // except 1d tensors
LLAMA_FTYPE_MOSTLY_Q1_3 = 37, // except 1d tensors
LLAMA_FTYPE_MOSTLY_TQ1_0 = 36, // except 1d tensors
LLAMA_FTYPE_MOSTLY_TQ2_0 = 37, // except 1d tensors
LLAMA_FTYPE_MOSTLY_Q1_3 = 38, // except 1d tensors
LLAMA_FTYPE_MOSTLY_Q2_2 = 39, // except 1d tensors
LLAMA_FTYPE_GUESSED = 1024, // not specified in the model file
};